A tumour-derived organoid biobank maps cancer gene dependencies

The multi-omics characterization of hundreds of 2D cancer cell lines and their perturbation in pharmacological and genetic screens^{2,3,4,5,6,7,8,9} has been used to build reference maps of cancer dependencies to inform precision cancer medicines. However, the widely available set of around 1,000 human cancer cell lines⁹ has limitations, including biased representation of certain cancer subtypes or absence of others, no patient-matched germline genetic data for accurately calling somatic variation, and a lack of linked patient characteristics or clinical data. Furthermore, most cell lines adapt to in vitro culture in ways that are not well understood, and reference patient-matched tumour samples are unavailable for comparisons, collectively hampering their use for pre-clinical research and discovery. Patient-derived xenograft models capture elements of the tumour microenvironment, but require specialized facilities and substantial resources, limiting broader use.

Tumour organoids are 3D cultures derived from patient tissue grown with niche factors in an extracellular matrix. They can be derived with higher success rates, capture inter-tumour heterogeneity and better reflect tumour histological, genetic and functional features than cell lines (reviewed in ref. ¹). However, scientific, technical and practical constraints have limited their widespread adoption¹⁰. Existing biobanks are relatively small (the two largest academic collections include 115 and 96 models), limiting representation of tumour diversity and so do not capture tumour heterogeneity^11,12. Many lack comprehensive genomic annotation, with only limited or partial molecular characterization, restricting links to clinical data and systematic evaluation of their fidelity to parent tumours. Some organoids are non-renewable or short-term cultures not suitable for biobanking^13,14, and others have restrictive ethical approvals that limit access. Finally, it is unclear whether a large, heterogeneous biobank of tumour organoids, given their greater complexity in culture compared to cell lines, can support systematic mapping of cancer vulnerabilities.

Here we report results from an extensive enterprise in the UK, in partnership with the Human Cancer Model Initiative (HCMI), to derive a highly annotated renewable tumour organoid biobank as an open community resource, and demonstrate its utility for mapping cancer dependencies to inform precision cancer medicine.

To create an organoid biobank, we established a network of five clinical sites in Birmingham, Cambridge, Glasgow, London and Southampton to access patient clinical data and obtain tumour material through surgical resections or biopsies, together with a germline reference normal sample (Fig. 1a and Methods, ‘Ethical approval and sample collection’). We obtained consent to enable broad sharing of models.

Fresh tumour samples were sent to the Wellcome Sanger Institute (May 2016 to November 2023) for organoid derivation in a single centralized facility using standardized operating procedures to minimize variability. A small number of derivations used flash-frozen tissue owing to shutdowns during the COVID-19 pandemic. A parallel sample from the same tumour specimen was acquired for whole-genome sequencing (WGS), enabling later analysis of model fidelity. We focused on five cancer types of unmet clinical need for which organoid derivation protocols were established, in decreasing order of sample numbers received: colorectal adenocarcinomas (COAD/READ), oesophageal adenocarcinomas (ESCA), ovarian carcinomas (OV), pancreatic adenocarcinomas (PAAD) and stomach adenocarcinomas (STAD). We applied stringent criteria for derivation success, requiring expansion to more than 25 million cells to create a bank of 25 cryovials, including the ability to cryopreserve and re-culture organoids (freeze–thaw quality control), and confirmation that their single nucleotide polymorphisms (SNPs) matched the original tumour. An organoid sample was prepared for molecular characterization during banking and freeze–thaw quality control.

Overall, organoid cultures were successfully established for 256 out of 907 samples from 878 unique donors (Supplementary Table 1), yielding an overall efficiency rate of 28% (Fig. 1b). The number of organoids in our biobank exceeds the number of commonly available cell lines for COAD/READ (132 compared with 90 in the Cancer Dependency Map⁹) and includes the cancer types ESCA¹⁵ and OV¹⁶, for which there are few existing representative models. Rare tubulovillous and colon tubular adenomas, and ovarian clear cell and mixed cell carcinoma subtypes are also represented in our biobank (Supplementary Table 2).

Derivation at a single facility enabled assessment of parameters associated with success (Extended Data Fig. 1 and Supplementary Table 1). The most common reasons for model failure were the development of non-renewable short-term cultures and insufficient tumour cellularity (labelled ‘lack of cells’; Extended Data Fig. 1a). When we account for these factors, by including short-term cultures and excluding samples with low cellularity, our success rate was 65%, comparable to previous studies. Success rates were similar between fresh and frozen tissue (Extended Data Fig. 1b). Once banked, 93% of cultures passed a freeze–thaw quality control step (Fig. 1b), supporting the effectiveness of our protocols to generate renewable cultures. Early passaging and cryopreservation (days to passage: P = 0.0167; days to bank: P = 0.0072; passages to bank: P = 0.0228; Wilcoxon rank-sum test) were unexpectedly associated with banking failure, potentially owing to the dominant outgrowth of non-tumour cells (Extended Data Fig. 1c–e).

A subset of 133 models was derived as part of the HCMI (Fig. 1c and Extended Data Fig. 2a), and these samples were independently genetically analysed in an accompanying HCMI paper¹⁷ (Extended Data Fig. 2b). We observed high concordance of somatic variants from these independent analyses (Extended Data Fig. 2c,d), with some discrepancy for detection of insertion–deletion mutations (indels), largely attributable to differences in sequencing coverage (Extended Data Fig. 2e; average coverage: Sanger, 38×; HCMI, 19×). These findings highlight the robustness and intercompatibility of results from both studies.

Thus we have developed one of the most extensive biobanks of long-term renewable patient-derived 3D organoids across five cancer types. Organoids are being distributed through non-profit and commercial repositories to enable open sustainable community access (Fig. 1c and Methods, ‘Availability of Organoid Biobank’). An inventory of all derived models, their unique identifiers and molecular and functional characterization is provided (Supplementary Table 2).

The lack of patient and clinical data for most widely used cancer cell lines limits their utility. Here, we acquired patient and clinical annotations for each model (ranging from 25 to 77, depending on cancer type and patient; Supplementary Table 2) (Fig. 1d). COAD/READ and ESCA organoid cultures were mostly from male patients, consistent with incidence rates^18,19. Our biobank includes organoids from patients with early (stage I and II; n = 83) and advanced (stage III and IV; n = 131) pathological staging (Fig. 1d). The biobank was derived from primary tumours (n = 212) and metastatic lesions (n = 44) (Fig. 1d), providing a valuable resource to study metastatic disease biology. Some patients had received prior treatment (n = 122), which was predominantly chemotherapy (n = 119), consistent with standard-of-care treatment, with 8 patients also receiving targeted immunotherapy, enabling the study of treatment resistance (Fig. 1d). Nearly 60% of patients with ESCA were smokers or ex-smokers, a known risk factor²⁰ (Fig. 1d), and patients undergoing surveillance for Barrett’s oesophagus had earlier stage ESCA (Extended Data Fig. 1f, P = 0.0375, chi-square test). A total of 17 organoids were derived from patients less than 50 years of age (Fig. 1d), including 11 with COAD/READ. Early-onset colorectal cancer is a leading cause of cancer death in young adults, with increasing incidence rates, underscoring the value of these models for studying this emerging clinical challenge²¹. Organoids from patients with advanced stage COAD/READ were more frequent on the proximal site (Extended Data Fig. 1g, P = 0.1761, Fisher’s exact test), as were tumours present in older patients (over 70 years old), consistent with clinical presentation²² (Extended Data Fig. 1h, P = 0.0165, Fisher’s exact test).

The biobank includes 13 patient-matched organoids from 6 patients, including 4 pre- and post-treatment paired samples, 3 organoids from one patient’s primary tumour and 2 sequential liver metastases, and 2 organoids from one patient’s matched liver metastases collected 1 year apart (Extended Data Fig. 3a).

Overall, the clinical information associated with each organoid—including disease stage, risk factors and prior treatment—enhance the relevance of the biobank, enable rational model selection and support integration with external clinical datasets for deeper interpretation.

To further inform the usefulness of our biobank, we performed a comprehensive genomic characterization of organoids and their patient-matched tumour sample using WGS, as well as RNA sequencing (RNA-seq) of the organoid (Fig. 1c). Single nucleotide variants (SNVs), indels, structural variants (SVs), somatic copy number alterations (SCNAs), mutational signatures, chromothripsis and whole-genome duplication (WGD) events were identified. WGS is available for all 256 organoids (Supplementary Table 2), together with the corresponding normal tissue sample as a germline reference to enable accurate calling of somatic variation, overcoming a major limitation of most 2D cell lines. For 171 organoids, sufficient tumour tissue was available for sequencing (Fig. 1c and Supplementary Table 2), enabling a direct comparison of organoid to tumour.

To assess model fidelity, we compared organoid cultures with their patient-matched tumour samples across multiple genomic features (Supplementary Table 3). Given the substantial genomic and clinical differences between COAD/READ samples with microsatellite instability (MSI) or microsatellite stability (MSS), we analysed these separately. The correlation for mutational load was high (Pearson correlation coefficient (PCC) = 0.786; Fig. 2a), as was the correlation for SVs (PCC = 0.823; Fig. 2b). The clonal fractions, which denote the proportion of mutations within a cancer cell fraction (CCF) above 0.75, were well correlated but higher in organoids (PCC = 0.92; Fig. 2c), indicating that most mutations are clonal and that intratumour heterogeneity is conserved. Concordance was also strong at the SNV and indel level, with 76 pairs (44%) sharing at least 75% of somatic mutations (Fig. 2d). Genome-wide SCNA profiles also showed high correlation (median PCC = 0.78), with 142 pairs (83%) achieving a PCC of at least 0.5. Correlations were lower in COAD/READ with MSI, a subtype that was not driven by SCNAs (Fig. 2e). For ploidy comparisons, 139 pairs (81%) exhibited ploidy differences within one standard deviation (s.d. = 0.697; Extended Data Fig. 4a). High SCNA correlation or an increased proportion of shared SNVs and indels between organoids and matched tumours was positively associated with higher tumour purity (Extended Data Fig. 4b,c), suggesting that concordance could be underestimated when tumour purity is low. Chromothripsis and WGD were more frequently detected in organoids than in tumours, and concordance between matched samples increased with tumour purity (Extended Data Fig. 4d–f). Finally, we compared patient-matched organoids and found them to be largely comparable, with some minor differences in SNVs and high conservation of cancer driver events (Extended Data Fig. 3b).

Together, these findings confirm that our organoids faithfully recapitulate a wide range of genomic features from the tumour from which they were derived, supporting their relevance for disease modelling. The higher tumour cellularity of organoids and potentially reduced heterogeneity due to clonal pruning during culturing lead to greater sensitivity to detect genomic alterations in organoid cultures than in tumours.

Having confirmed the fidelity of organoid cultures to their original tumour, we next examined in detail these genomic features of the organoids by tumour type (excluding STAD organoids owing to low numbers; Supplementary Table 3). Almost all organoids had more than 0.95 tumour purity, consistent with a highly enriched tumour cell culture (Fig. 3a). Most COAD/READ-MSI models were diploid as expected²³, whereas other cancer types had greater ploidy variation²⁴ (median = 2.5; Fig. 3a). MSI models had a high mutational load (Fig. 3a). ESCA and OV organoids had the highest number of SVs (Fig. 3a). Clonal fractions were overall high across the models, especially for COAD/READ-MSI organoids, indicating a significant degree of clonality. However, organoids with clonal fractions below 50% may exhibit intra-organoid heterogeneity (Fig. 3a).

Lineage-specific cancer driver events reported in public cancer datasets²⁵ were recapitulated in the organoid biobank at comparable frequencies (Fig. 3b and Extended Data Figs. 5 and 6a). Complex rearrangements and SCNAs were observed with tumour-type-specific patterns consistent with patient tumours (Fig. 3b). COAD/READ-MSI models had almost no complex rearrangements, whereas OV organoids had the highest frequency of WGD and chromothripsis²⁶.

A key feature of our large biobank is representation of inter-patient heterogeneity, including rare but clinically important genotypes that are underrepresented or absent in cancer cell line collections⁹ (Extended Data Fig. 5). Our biobank also includes rare models with complex genotypes, such as five COAD/READ-MSS models that all carry APC, TP53, KRAS and SMAD4 mutations or one with APC, TP53, KRAS and PIK3CA mutations together with a PTEN deletion, which will enable studies investigating acquisition of mutations during multi-step carcinogenesis^27,28. Our biobank includes 94 KRAS-altered organoids, including specific variant alleles that varied in frequency by tumour type, consistent with patient tumours²⁹, thereby enabling the study of this clinically important molecular cohort (Extended Data Fig. 5).

We used our WGS data to uncover the landscape of mutational signatures available in the Catalog of Somatic Mutations in Cancer (COSMIC) database³⁰ in organoids, providing insights into disease aetiology and patient clinical history (Extended Data Fig. 6b and Supplementary Table 3). For example, all MSI models had signatures related to DNA mismatch repair deficiency (SBS15, SBS44, SBS26, SBS20, SBS6 and/or the recently described SBS_MSI_M). SBS88, caused by the bacteria-produced mutagen colibactin^31,32, was detected in 10 COAD/READ-MSS organoids, including those from three early-onset cases, representing 27% of samples from patients with COAD/READ who were diagnosed before the age of 50. Notably, 9 out of 10 SBS88-positive organoids originated from distal (n = 4) or rectal (n = 5) tumours³³. We identified six organoids with SBS3, a signature linked to homologous recombination deficiency (HRD), some of which overlapped with models detected as HRD by CHORD, which integrates multiple types of somatic mutations to detect HRD³⁴ (Extended Data Fig. 5). We detected SBS35, associated with previous platinum treatment, in organoids from donors treated with this chemotherapeutic agent (Extended Data Fig. 3c). The high mutational load of cancer cell lines, coupled with the absence of matched normal samples for germline filtering, complicates accurate detection of mutational signatures³⁵. By contrast, organoid cultures offer a promising model for studying signature biology, capturing aspects of disease aetiology including genetic factors and clinically relevant exposures.

RNA-seq of the organoids was performed, enabling us to classify the COAD/READ models into two of the most validated and clinically relevant transcriptional subtypes: consensus molecular subtypes³⁶ (CMS 1–4) and colorectal cancer intrinsic subtypes (CRIS A–E)³⁷ (Supplementary Table 3). CMS and CRIS classifications showed strong concordance in colorectal cancer organoids and were associated with distinct genomic and anatomical features, with CRIS providing more precise correlations with specific molecular alterations (Extended Data Fig. 6c).

Whole-genome and transcriptomic profiling showed that organoids maintained stable ploidy, mutational landscapes and transcriptional profiles over up to six months of culture and multiple freeze–thaw cycles (Extended Data Fig. 7 and Supplementary Information). Long-term culture caused slightly greater—but still modest—genetic drift than repeated freeze–thaw cycles. These results demonstrate that organoids largely preserve genomic integrity over time^38,39, supporting their use as robust and renewable tumour models.

These comprehensive genomic and transcriptomic analyses underscore the utility of organoids in reflecting inter-tumour heterogeneity, lineage-specific driver mutations, mutational patterns and transcriptional programs associated with disease and that can be linked to clinical parameters. This study is one of the most comprehensive genomic analyses of organoids performed to date, both in terms of number of models and depth of characterization. To facilitate accessibility of these data, we have made them available through the Cell Model Passports database⁴⁰ (https://cellmodelpassports.sanger.ac.uk/).

Dependency catalogues from pharmacological and genetic screens, including the Cancer Dependency Map, have advanced target discovery and patient stratification. Yet existing cell models do not capture the full tumour diversity (for example, in ESCA), and inter-tumour heterogeneity limits genotype–dependency associations. High mutational burden and lack of matched germline data further complicate linking dependencies to genomic features. Systematic dependency mapping in organoids could address these gaps. However, CRISPR screens in 3D cultures remain limited^41,42,43,44, as the increased complexity of screening in 3D cultures across tissue histologies is constrained in scale. To overcome this, we established a scalable platform to map dependencies across our organoid biobank.

Organoids were screened using a 5% basement membrane extract-2 (BME-2) suspension culture, which simplifies handling while preserving genetic and phenotypic stability³⁸. Screening was conducted using the minimal genome-wide CRISPR–Cas9 library⁴¹ (MinLibCas9), alongside the larger Human CRISPR Library Yusa v.1.1 (refs. ^3,9) (hereafter, Yusa v1.1) for a subset of models (Extended Data Fig. 8a). MinLibCas9 required fewer cells than Yusa v1.1 and, combined with suspension culture, increased efficiency of dependency mapping in organoids to enable systematic screening.

We completed quality-control-passed CRISPR–Cas9 screens on 162 unique organoid cultures from 5 tumour types (85 COAD/READ (exceeding the 63 cell lines screened in DepMap), 59 ESCA, 11 OV, 4 PAAD and 3 STAD; overall 85% success rate), using both MinLibCas9 and Yusa v1.1 libraries for 16 models (Extended Data Fig. 8 for screen quality control and Supplementary Table 4). Using the BAGEL2 method⁴⁵, we identified a median of 1,440 fitness genes in each organoid (range 297–2,160; Fig. 4a and Supplementary Table 5), fewer than reported in cell lines (median = 2,507 for 930 cell lines⁹). The number of fitness genes was inversely proportional to the log fold change (LFC) standard deviation and other quality control metrics (Fig. 4b, Extended Data Fig. 8g,h and Supplementary Tables 4 and 6), indicating that greater variability in gene depletion negatively affects fitness gene identification. We applied rigorous quality control criteria to screens, including replicate concordance, clear discrimination of reference essential versus non-essential genes (area under receiver operating characteristic curve (AUROC) = 0.97 and area under precision-recall curve (AUPR) = 0.973) and good effect size (Fig. 4c,d and Supplementary Table 4), with results comparable to cell lines^3,9 (AUROC = 0.92 and AUPR = 0.9). Our results demonstrate the feasibility of constructing organoid dependency maps, with the potential to uncover new biology.

We applied the Adaptive Daisy Model (ADaM) to define core fitness genes in organoids, and compared these with those in cell lines⁹. This identified 654 shared core fitness genes and 97 that were unique to organoids (Supplementary Table 4). Gene set enrichment analysis of organoid-specific core fitness genes highlighted strong associations with steroid biosynthesis (most significant: Biological Process—steroid biosynthetic process, P value = 1.38 × 10⁻⁶; most enriched: Biological Process—isopentenyl diphosphate biosynthetic process, enrichment fold change = 667.5, Fisher’s exact test), potentially linked with a role for cholesterol in regulating intestinal stem cells and tumourgenesis⁴⁶ (Extended Data Fig. 9a and Supplementary Table 4).

Genetic constraint, derived from population-scale human genome sequencing, reflects the degree to which a gene is intolerant to heterozygous variation and provides an orthogonal method to identify disease-linked or functionally important (essential) genes^47,48. Core fitness genes experimentally derived from CRISPR–Cas9 screens in cell lines⁹ and organoids (this study) were enriched for genetically constrained genes (high categories), as were reference lists of essential genes (Fig. 4e). By additionally excluding genes commonly essential in cell lines (area under the curve (AUC) method), we refined our list to include 18 of 97 organoid-specific core fitness genes and annotated them with mouse knockout phenotypes and GTEx tissue-specific expression (Extended Data Fig. 9b). Genes with low constraint and normal mouse knockout phenotype, such as isopentenyl diphosphate delta isomerase 1 (IDI1), required for isoprenoid synthesis and involved in cholesterol production, may reflect cancer organoid selective fitness dependencies. Conversely, genes under high genetic constraint and with severe knockout phenotypes represent candidate core fitness genes identified in organoids—for example, ITGB1 (Extended Data Fig. 9b,c). Together, this integrative approach combining experimental data from organoids with genetic constraint from human populations identifies core fitness genes.

The number of gastrointestinal tract (COAD/READ, ESCA, PAAD and STAD, n = 151) and specifically COAD/READ (n = 85) and ESCA (n = 59) organoids screened with CRISPR, and their genomic deep characterization, enabled analyses to interpret differential gene dependencies. To focus on context-specific effects, we excluded core fitness genes, curated essential and non-essential genes, non-expressed genes, and genes whose depletion was restricted to or absent from a single organoid. For the remaining genes, we performed differential dependency analysis comparing LFC values between organoids where a gene was essential or non-essential based on BAGEL2 scores (minimum of two organoids per group). This approach identified 7,086, 5,103 and 4,082 differentially dependent genes in gastrointestinal, COAD/READ and ESCA organoids, respectively (Fisher’s exact test, adjusted P value < 0.05; Supplementary Table 7).

Annotation by genetic constraint showed a distribution across cancer types, in contrast to essential genes (enriched for high constraint) and non-essential genes (enriched for low constraint) (Fig. 4e). However, within these differential dependencies, genes with higher genetic constraint tended towards stronger fitness effects (lower mean LFC), as illustrated in gastrointestinal organoids (Fig. 4f). These observations reveal a broad landscape of differential gene dependencies in organoids and constrained genes tend to show greater dependency penetrance across organoids.

To gain insights into the underlying mechanisms of these differential dependencies, we used linear regression models to link gene fitness effects with our curated genomic, patient and clinical features. From this analysis, we identified 1,841 significant gene–biomarker associations (based on false discovery rate (FDR)-adjusted P values and effect size; Extended Data Fig. 9d and Supplementary Table 8), distributed across categories of genetic constraint (Fig. 4f, labelled A, B or C depending on strength of statistical association). In COAD/READ, we observed dependency on WRN associated with a RNF43 mutation (effect size = −0.747, adjusted P value = 0.01), RPL22 or SETD1B mutations (effect size = −0.641, adjusted P value = 0.027) and mutational signature SBS44 (effect size = −0.823, adjusted P value = 0.0009), consistent with MSI status due to deficient DNA mismatch repair (Extended Data Fig. 9e) and showcasing the value of integrating multi-omics biomarkers. Notably, four MSI organoids were not WRN-dependent, and lacked expanded TA dinucleotide repeats associated with deficient DNA mismatch repair and underlying WRN dependency and inhibitor sensitivity⁴⁹, showing the utility of a large biobank to sub-stratify responses.

We also identified MDM2 dependency for TP53-wild-type (effect size = 0.463, adjusted P value = 0.001) and unexpectedly for KRAS^G12D COAD/READ organoids (effect size = −0.576, adjusted P value = 0.01), of which 4 out of 5 were TP53-wild-type. HRAS gain-of-function events (effect size = −0.829, adjusted P value = 0.0019) partially mirrored MDM2 gain-of-function events (effect size = −0.566, adjusted P value = 0.0039), and all of them correlating with sensitivity to nutlin-3a (an MDM2 inhibitor, PCC = 0.49) (Extended Data Fig. 9f). For ESCA, CCNE1 amplification, which occurs in 7% of patients⁵⁰, was associated with dependency on CCNE1 (effect size = −0.645, adjusted P value = 3.65 × 10⁻⁵) (Extended Data Fig. 9g). Across gastrointestinal organoids, INTS6L expression correlates with INTS6 dependency (effect size = 2.229, adjusted P value = 5.28 × 10⁻¹⁵) (Extended Data Fig. 9h), as previously described in cell lines⁵¹. This rich landscape of dependency–biomarker associations illustrates the value of integrating genomic, transcriptomic and clinical features, and highlights novel hypotheses for functional investigation. Some biomarker-linked differential dependencies map to genes under low genetic constraint (Fig. 4f), which may render them suitable candidate therapeutic targets.

The EGFR–RAS–MAPK signalling pathway is frequently dysregulated in COAD/READ and is targeted by US Food and Drug Administration (FDA)-approved inhibitors of EGFR and KRAS, with an expanding pipeline of RAS agents in development that differ in variant allele specificity, isoform selectivity and mechanism of action⁵². Yet responses remain variable, highlighting the need to define the biological contexts in which specific KRAS inhibitors will confer the greatest therapeutic benefit. Our biobank provides a platform to tackle these questions. In our cohort of 85 COAD/READ organoids with CRISPR data, dependency on this pathway was heterogeneous. KRAS-mutant organoids had significantly greater dependency on KRAS than KRAS wild-type organoids, whereas KRAS wild-type models had greater dependency on EGFR and PTPN11, influenced by alterations in other pathway components such as BRAF (Fig. 5a). Notably, KRAS dependency varied by allele, with G12X variant organoids (in which X denotes any amino acid substition) having the greatest dependency (Fig. 5a). Compared with other KRAS-mutant organoids, G12X models also showed increased EGFR and PTPN11 dependency.

To further investigate EGFR–RAS–MAPK pathway dependencies, we profiled drug sensitivity in a subset of KRAS-mutant (n = 17) and KRAS wild-type (n = 4) colorectal organoids, selected to represent diverse KRAS alleles and dependency profiles (Extended Data Fig. 10a and Supplementary Table 9). Given the diversity of the cohort, we tested pan-RAS, KRAS-specific and variant-specific inhibitors, as well as an EGFR inhibitor and a dual EGFR–ERBB2 inhibitor. All organoids were highly sensitive to the pan-RAS(ON) inhibitors RMC-6236 (mean half-maximal inhibitory concentration (IC₅₀) = 0.042 μM) and RMC-7977 (mean IC₅₀ = 0.023 μM) (PCC = 0.89 for organoid IC₅₀ values; Extended Data Fig. 10b). By contrast, the activities of the pan-KRAS(OFF) inhibitor BI-2865 and the KRAS degrader ACBI3 were similar, but showed overall lower potency (mean IC₅₀ ≥ 10 μM for both; PCC = 0.86 for organoid IC₅₀ values; Extended Data Fig. 10c). The G12C allele-specific inhibitor sotorasib selectively targeted KRAS^G12C organoids (P value = 0.044; Extended Data Fig. 10d), whereas MRTX1133 (G12D allele-selective) inhibited G12D but also affected non-G12D organoids (Extended Data Fig. 10e). Overall, we observe variable sensitivity to RAS inhibitors with different modes of action.

Organoids with high CRISPR-derived EGFR dependency were amongst the most sensitive to the EGFR inhibitors afatinib and gefitinib. Notably, KRAS wild-type and KRAS^G12X organoids had greater sensitivity compared to KRAS non-G12 variants (Fig. 5b). Upon combined EGFR and KRAS inhibition, only a subset of KRAS^G12X-mutant organoids exhibited pronounced loss of viability (combined maximal effect (E_max) > 85%) and synergy (Bliss > 0.25), including when using the pan-KRAS(OFF) inhibitor BI-2865 (Syn/Eff in Extended Data Fig. 10a,f), in part owing to potent monotherapy activities. Finally, a KRAS^Q61H variant organoid was unresponsive to EGFR inhibition by either CRISPR or small molecules (Fig. 5b). We further confirmed this by showing that EGF withdrawal from the culture medium did not reduce viability in Q61H organoids, unlike in G12X or wild-type organoids (Fig. 5c). Collectively, the biobank and dependency map yield clinically relevant functional insights into the EGFR–KRAS–MAPK pathway, revealing substantial heterogeneity in responses to RAS and EGFR inhibition and showing that EGFR dependency varies across organoids with distinct KRAS variant alleles⁵³.

Derivation and interrogation of patient-matched samples pre- and post-treatment can inform on tumour evolution, mechanisms of resistance and approaches to target treatment-refractory cells. Patients with ESCA have a poor prognosis and few treatment options, particularly in advanced disease. We therefore compared the CRISPR–Cas9 dependency profiles for paired ESCA organoids from a 69-year-old man with stage III disease derived from a laparoscopic biopsy (HCM-SANG-0299-C15) before chemotherapy (epirubicin plus capecitabine plus cisplatin) and following surgical resection at the time of relapse (HCM-SANG-0299-C15-B). Both organoids retained a KRAS amplification, whereas a MYC amplification in the pre-treatment was absent in the post-treatment organoid (Extended Data Fig. 3b), implying clonal evolution under the selective pressure of chemotherapy. Although the overall dependency profiles of the matched organoids were similar, MYC dependency was reduced in the post-treatment organoid, consistent with loss of MYC amplification (Fig. 5d).

Curation of differential dependencies post-treatment identified several druggable targets, including a reduced dependency on KRAS and DNMT1 and increased dependency on the proteasome core subunit PSMB5 (Fig. 5d), which we validated experimentally. The post-treatment organoid had reduced sensitivity to two pan-RAS inhibitors (RMC-6236 and RMC-7977) and two DNMT1 inhibitors (5-azacytidine and decitabine), and increased sensitivity to two proteosome inhibitors (bortezomib and ixazomib), consistent with predictions from CRISPR experiments (Fig. 5e). There was no difference in TOP2A dependency (Fig. 5d) or sensitivity to the TOP2A inhibitor etoposide in the matched samples (Extended Data Fig. 10g), indicating that differential effects are likely to be target-specific. The organoid derived from the pre-treatment sample also exhibited increased sensitivity to epirubicin and cisplatin versus the post-treatment sample, aligning with clinical relapse (Extended Data Fig. 10g). Our biobank captured the functional impact of treatment driven tumour evolution, selecting for cancer driver events and conferring differential dependencies that can be unmasked through CRISPR screening and that are candidates for drug repositioning.

Reductionist models are central to cancer biology and drug discovery. However, cancer heterogeneity strongly influences prognosis and treatment response, creating a need for larger, more representative model collections to advance precision medicine. Our comprehensive, highly annotated biobank complements 2D cell lines and is designed to catalyse discovery. To maximize access, models are deposited in public repositories and datasets are available via the Cell Model Passport⁴⁰ and DepMap Miner (https://dataminer.depmap.sanger.ac.uk/) portals.

This study is distinctive for the number of highly annotated organoids and the open availability of detailed genomic, patient and clinical data for each organoid¹⁰. Future biobanking would benefit from deeper tumour annotation, including histology, single-cell and spatial transcriptomics, and matched stromal and immune cultures⁵⁴. We used tissue-specific, standardized conditions to generate a molecularly diverse biobank and enable cross-model comparisons. Tailored culture conditions may enrich specific molecular subtypes and alter gene dependencies⁵⁵, and xenotransplantation provides an in vivo context to assess microenvironmental effects.

Genome-wide CRISPR screens in tumour organoids are rare^42,56, yet we map dependencies across a diverse biobank and achieved cell line-level quality despite the added complexity of 3D culture. Distinguishing cancer-selective dependencies from core fitness genes may inform therapeutic index, although matched normal cells are rarely available. To strengthen interpretation, we integrated population-scale genetic constraint data⁴⁸, providing orthogonal support for core fitness genes and indirect evidence of target tolerability.

Our organoid dependency map identified novel and clinically relevant vulnerabilities. We found allele-specific differences in KRAS and EGFR dependencies, with KRAS^G12X variants showing greater reliance on KRAS and upstream EGFR signalling. This supports a model^53,57 in which G12X tumours remain responsive to upstream receptor activation, unlike signalling-independent Q61X variants, with therapeutic implications for targeting KRAS^G12X cancers, and has been proposed to underlie the enrichment of KRAS^Q61X mutations as secondary resistance events to EGFR inhibitors in colorectal cancer. Additionally, by generating and dependency mapping patient-matched organoids before and after treatment, we traced tumour evolution under therapeutic pressure and uncovered new hypotheses for intervention in treatment-refractory ESCA. In summary, we provide a preview of a cancer dependency map across a diverse cohort of highly annotated organoid cultures, with potential to enhance and extend existing efforts as a tool for discovery.

Large-scale sequencing efforts^58,59 defined tumour mutational landscapes and transformed understanding of cancer biology. A next frontier in precision oncology is to move beyond descriptive genomics towards multi-scale, integrative functional maps that enable mechanistic modelling of disease, including virtual cell frameworks^60,61. This will require systematic modelling and perturbation across representative tumour cohorts integrated with patient data. Our organoid biobank and dependency map mark a major step towards building functionally informed maps of human cancer.

Ethical approval and sample collection

All necessary ethical approvals to conduct this work and to ensure that the derived organoids could be used for academic and commercial purposes were obtained from each participating clinical site for the collection of patient samples as well as for the Wellcome Sanger Institute for organoid derivation (IRAS ID:203519; REC 16/LO/1110). Further details are provided in Supplementary Information.

Availability of Organoid Biobank

Organoids have ethical approval for use in academic and commercial purposes. Models are being distributed by American Type Tissue Culture (ATCC) as part of the HCMI collection (https://www.atcc.org/hcmi) and EMD Millipore. Repository details and model availability are provided in Supplementary Table 2.

Organoid derivation

Detailed protocols for organoid derivation, cryopreservation and routine culturing are published on protocols.io and include embedded demonstration videos and workflow diagrams^62,63,64. Tumour samples were washed 3 times with PBS, minced, and either cryopreserved after centrifugation (800g, 2 min)⁶³ or enzymatically digested for 1–2 h at 37 °C. The suspension was filtered (100 μm), centrifuged and washed to remove debris and digestion buffer⁶². Isolated cells were embedded in approximately 15 μl droplets of extracellular matrix (80:20 basement membrane extract (BME):medium; 6.4–9.6 mg ml⁻¹ protein; Cultrex BME Type 2 Select 3532-001-02) and plated in pre-warmed 6-well plates following established protocols⁶⁴. After polymerization (15–20 min, 37 °C), 2 ml of organoid medium prepared using established recipes^39,65,66,67 was added, supplemented with antibiotics and 1 μl ml⁻¹ ROCK inhibitor (Y-27632, Staratech Scientific S1049-SEL-5mg).

After expansion to ≥25 million cells, 25 cryovials were banked and pellets collected for sequencing. Post-thaw viability was confirmed by re-culturing for four passages with freeze–thaw quality control assessment.

Organoid culture

Organoids were maintained either in 80% BME-2 droplets or in 5% BME-2 suspension culture³⁸. In the 80% BME-2 culture, organoids were cultured as described above. For the 5% suspension method, cancer organoids were suspended in a medium/extracellular matrix (ECM) dilution. For example, combining 10 ml of medium with 500 μl of BME-2 achieved the desired concentration, and organoids were then cultured in ultra-low-adherent flasks or plates.

For passaging, the medium, cells and ECM were collected and centrifuged at 800g for 2 min. After discarding the supernatant, organoids were dissociated using TrypLE (Gibco 12604013), an enzymatic reagent. The suspension was incubated for 10–60 min, allowing dissociation into small clumps. The cells were then pelleted and replated as described⁶⁴.

Model identifiers

All models presented in this study have a sanger_ID starting with ‘WTSI’, corresponding to the Wellcome Trust Sanger Institute. Additionally, if a model is shared with the HCMI, it also has an HCMI ID starting with ‘HCM’. When both IDs are available, the HCMI nomenclature is used as the sample_ID. The equivalences are shown in Supplementary Table 2.

Whole-genome sequencing and analysis

Sequencing and alignment

DNA extracted from snap-frozen tumour tissues, snap-frozen organoid cell pellets, blood or formalin-fixed paraffin-embedded normal tissues samples were prepared for WGS. Whole genome paired-end sequencing reads (150 bp) were generated using the Illumina HiSeq X Ten platform with an average coverage of 38×, comparable to coverage used for PCAWG. Reads were aligned using the BWA-MEM (v0.7.17) tool⁶⁸. PCR duplicates, unmapped and non-uniquely mapped reads were filtered out before downstream analysis.

SNV and indel calling

Somatic SNVs and short indels were identified using cgpCaVEMan⁶⁹ and cgpPindel⁷⁰, respectively. Germline variants and artefacts were filtered using matched normal samples and a panel of normals, with further post-processing using cgpCaVEManPostProcessing (https://github.com/cancerit). Variant allele frequencies (VAFs) were estimated using vafCorrect (v2.4.0)⁷¹, and variants with VAF > 0.05 were retained. Further details in Supplementary Information.

Structural variant and copy number calling

Copy number and allele-specific information were derived using AMBER (v3.5)⁷² and COBALT (v1.11)⁷². Somatic and germline SNVs and indels were identified using SAGE (v2.8) and SnpEff (v5.0)⁷³. Somatic SVs were called using GRIDSS2 (v2.12.0)⁷⁴, annotated with RepeatMasker (v4.1.2)⁷⁵ and kraken2 (v2.1.2)⁷⁶, and filtered with GRIPSS (v1.9)⁷⁷. Subsequently, this information was integrated to calculate the microsatellite status, tumour purity, ploidy, WGD and SCNAs using PURPLE (v2.54)⁷². SCNAs were converted into discrete copy number states. SAGE, GRIPSS, AMBER, COBALT and PURPLE were developed by the Hartwig Medical Foundation (HMF) (https://github.com/hartwigmedical/hmftools). Further details in Supplementary Information.

Cancer driver event annotation

We compiled a list of 783 cancer driver genes, derived as the union of two complementary gene sets from the IntOGen⁷⁸ and COSMIC⁷⁹ databases, as previously reported⁹. Each gene was annotated by its mechanism of action as either activating (Act, for oncogenes), loss-of-function (LoF, for tumour suppressor genes), ambiguous (with evidence of both mechanisms), or fusion. The complete list of cancer driver genes is in https://cellmodelpassports.sanger.ac.uk/downloads.

We collated individual putative cancer driver mutations (for example, frameshift, nonsense, stop-lost, exonic splicing silencer, missense or in-frame mutations) among SNVs and indels within these cancer driver genes from four data sources: IntOGen (including Cancer Genome Interpreter and BoostDM)⁷⁸, MSKCC⁸⁰, and cancer predisposition variants, that were identified by overlap with a reference set of pathogenic germline variants with matching effect⁸¹.

Loss of heterozygosity

We considered a tumour suppressor gene or an ambiguous gene to have loss of heterozygosity if the DNA copy number (CN) of the minor allele was <0.5 and the difference between the round ploidy and the round total CN was >0 (there was no amplification in the non-mutated allele) or if the VAF of a loss-of-function mutation was >0.85.

Biallelic alterations

Biallelic alterations included all cases with loss of heterozygosity, as well as homozygous deletions, SV disruptions or instances where each allele was affected by a different loss-of-function mutation.

Multiplicity, CCF and clonal mutations

SNVs were intersected with segment SCNAs using the GRanges and findOverlaps functions from the GenomicRanges R package (v1.56.1) to obtain information on the major allele, minor allele, and total tumour CN for each SNV, along with the VAF and tumour purity. The multiplicity (the number of chromosomal copies harbouring a given mutation) and CCF (the proportion of cancer cells carrying a specific mutation within a tumour sample) were then calculated according to the formulas in Steele et al.⁸² and Dentro et al.⁸³. Clonal mutations, defined as genetic alterations present in all cancer cells within a tumour, were identified as those with a CCF > 0.75.

CN correlation between paired organoids and tumours

SCNA segment data was divided into 100 kb bins across the genome, and the CN for each segment was calculated as the mean CN of all positions within the 100,000 bp window. Positions listed in the ENCODE blacklist⁸⁴ (https://github.com/Boyle-Lab/Blacklist) were excluded from this analysis. Pearson correlations were then computed using the mean CN for each segment between paired organoid and tumour samples.

Focal amplifications

Focal amplifications were defined as the presence of at least two genomic segments, each exceeding 100 kb in size, with a log₂(CN/ploidy) value > 6.

Signature analysis

Mutational signatures were extracted using SigProfilerExtractor (v1.2.2)⁸⁵ for the tumour and organoid samples from each tumour type separately. COAD/READ-MSI samples were analysed separately from COAD/READ-MSS samples. STAD samples were not analysed due to insufficient sample size. The resulting matrix file was used as an input for SigprofilerAssignment (v0.2.5)³⁰ to assign known COSMIC v3.4 signatures within single base substitution (SBS). De novo signatures were further refitted using the COMICv3.4 and an additional set of novel CRC and MSI specific signatures identified and validated in Mutographs project⁸⁶.

For data visualization, COSMIC mutational signatures representing less than 10% of mutations within each sample were categorized as ‘Others’ and excluded from calculations of both the proportion of models with that mutational signature and the proportion of mutations representing each signature by cancer type. SBS5 and SBS40a were collapsed into a single category, based on the hypothesis that these signatures arise from a combination of correlated mutational processes^87,88.

Homologous recombination deficiency

Homologous recombination deficiency was assessed using CHORD (v2.0.3)³⁴.

Complex genomic rearrangements

Chromothripsis and other complex genomic rearrangements were detected using ShatterSeek (v1.1)²⁶, as previously described⁸⁹.

RNA-seq and analysis

Paired-end transcriptome reads (75 bp) were quality filtered and mapped to GRCh38 (ensemble build 98) using STAR (v2.5.0c)⁹⁰ with a standard set of parameters (https://github.com/cancerit/cgpRna). Resulting bam files were processed to get the per gene read count and transcripts per million (TPM) data using RSEM (v1.3.3)⁹¹. TPM values were used for the downstream analysis.

CMS/CRIS subtypes

Consensus molecular subtypes (CMS) and colorectal cancer intrinsic subtypes (CRIS) subtypes were inferred for COAD/READ organoids with CMSCaller^36,92 using RSEM expected count data.

CRISPR–Cas9 screening

Detailed protocols including process diagrams and example data for generating Cas9-expressing organoid cultures and CRISPR–Cas9 library transduction in organoid are published on protocols.io^93,94. Stable Cas9-expressing organoids were generated using lentiCas9-Blast (Addgene 52962) with polybrene (8 μg ml⁻¹). Following overnight incubation, medium was replaced with complete medium containing Y-27632 (2.5 μM), and blasticidin (Invivogen, ant-bl-1, 10 mg ml⁻¹) selection was applied 120 h after transduction. Cas9 activity was measured using a fluorescent reporter assay (Addgene 67982 and 67981)⁹⁵. Only organoid lines with activity of 75% or more were selected for single guide RNA (sgRNA) library transduction.

Two genome-wide CRISPR–Cas9 sgRNA libraries were used. The Human CRISPR Library Yusa v.1.1 (ref. ³), containing 100,086 sgRNAs that target 18,009 genes (with 5–10 sgRNAs per gene) and 1,004 non-targeting sgRNAs; and the Minimal Genome-Wide Human CRISPR–Cas9 Library (MinLibCas9, Addgene 164896)⁴¹, which includes a selected sgRNA primarily drawn from Yusa v1.1 comprising 37,522 sgRNAs that target 18,761 genes (with 2 optimal sgRNAs per gene) and 200 non-targeting sgRNAs (Extended Data Fig. 8a). Lentiviral volume required for multiplicity of infection of 0.3 was determined by titration, and transduction efficiency assessed by BFP flow cytometry.

A total of 12.5 × 10⁸ (Yusa v1.1) or 3.3 × 10⁷ (MinLibCas9) cells were transduced in triplicate using the same batch of Cas9-transduced organoids. No significant batch effects were observed across independent batches of Cas9-expressing organoid lines (Extended Data Fig. 8b). Cells were infected with the lentiviral-packaged whole-genome sgRNA library (for 100× coverage), in medium containing polybrene (8 μg ml⁻¹), and Y-27632 (2.5 μM). Following overnight incubation, cells were plated in fresh medium in a 5% suspension. Transduction efficiency (target of 30%) was confirmed on day 6. Screens were maintained under puromycin selection (Invivogen, ant-pr-1, 10 mg ml⁻¹) for a further 16 days (21 days in screen in total). A final selection efficiency of at least 60% was required for the screen to pass. At the end of the screen approximately 2.5 × 10⁷ cells were collected, pelleted, and stored at −80 °C for downstream processing.

CRISPR screen data processing

CRISPR screens performed using the Yusa v1.1 (ref. ³) and MinLibCas9 (ref. ⁴¹) libraries were harmonized by restricting analyses to shared sgRNAs, enabling integrated downstream processing. Quality control procedures, adapted from established cell line screening pipelines³, were applied to assess replicate concordance, sgRNA representation and classifier performance in distinguishing essential from non-essential genes. Low-quality replicates and organoids failing predefined quality control criteria were excluded prior to further analysis. Read counts were normalized and corrected for copy-number effects using CRISPRcleanR (v3.0.1)⁹⁶, followed by batch correction across libraries⁹⁷. Gene-level fitness effects were estimated using BAGEL (v2)⁴⁵ with curated reference gene sets, and LFC values were scaled relative to essential and non-essential controls to facilitate cross-organoid comparability. Full details of sgRNA selection, quality control thresholds, normalization, batch correction, statistical procedures and final selection of models are provided in the Supplementary Information.

CRISPR screen processed data analysis

Analysis of organoid core fitness genes

We applied ADaM implemented in the CoRe R package (v1.0.0) (https://github.com/DepMap-Analytics/CoRe)⁹⁸ using the same curated reference essential gene set as previously mentioned³ to calculate false-positive rates. This analysis identified 751 core essential genes (Supplementary table 4). Although we used all organoids together as input, only three cancer types were included (PAAD and STAD organoids were excluded), so the resulting core fitness genes do not represent a pan-cancer set.

We compared core fitness genes in organoids with those identified in cell lines⁹ and observed an overlap of 654 genes, with 97 genes identified as organoid-specific (Extended Data Fig. 9). We performed a Fisher’s exact test on these organoid-specific core fitness genes to perform pathway enrichment analysis, using Gene Ontology Biological Processes, KEGG pathways⁹⁹ and Hallmarks¹⁰⁰ obtained from the msigdbr R package (v7.5.1)¹⁰¹. For this analysis, PanCancer common essential genes identified using the AUC method were excluded⁹⁸. For visualization, when multiple pathways shared the same set of organoid-specific core fitness genes, only the pathway with the most significant P value was displayed.

Differential dependency analysis in all gastrointestinal, COAD/READ and ESCA organoids

We conducted an analysis to identify genes with the greatest dependency variability across organoids within each cancer type and to highlight context-specific dependencies. The analysis proceeded as follows:

1. 1.

   Gene filtering based on consistent depletion/non-depletion. Using binary dependency matrices generated by BAGEL2 for each cancer type, we excluded genes that were only considered depleted in a single organoid and not depleted in only one organoid.

2. 2.

   Exclusion of core fitness and control genes. Genes previously classified as pan-cancer core fitness genes in cell line datasets, the 751 organoid-specific core fitness genes identified in this study, as well as control sets of essential and non-essential genes, were removed to ensure a focus on genes with differential dependency profiles.

3. 3.

   Expression thresholding. Genes with a mean expression level of log₂(TPM + 1) < 0.1 in each cancer type (considered ‘not expressed’) were also excluded from the analysis.

Following these filtering steps, for each remaining gene, we calculated the difference in LFCs (delta LFC) between organoids where the gene was depleted and those where it was not and tested the statistical significance with a Fisher’s exact test. Only genes with a FDR-adjusted P value < 0.05 were considered differentially dependent (7,086 for gastrointestinal, 5,103 for COAD/READ and 4,082 for ESCA; Supplementary Table 7).

Biomarker analysis

To identify molecular and clinical features associated with context-specific gene dependencies, we performed a systematic biomarker analysis across gastrointestinal, COAD/READ and ESCA organoids. Candidate dependencies were selected from the differential dependency analysis based on recurrence criteria and evaluated against a curated set of genomic, transcriptomic and clinical features, including driver mutations and specific variants, copy number alterations, structural events, mutational signatures, gene expression, pathway activity scores, and composite loss- and gain-of-function events. Associations between gene fitness effects and features were tested using linear regression models incorporating relevant technical and biological covariates, with significance assessed using likelihood-ratio tests and multiple testing correction. Significant associations were prioritized based on adjusted P value and effect size and classified into tiers reflecting strength of association. Full details of feature selection, model specification, statistical thresholds and classification criteria are provided in the Supplementary Information.

High-throughput drug screens

For high-throughput screening, organoids were dissociated into single cells, counted and seeded in 5% BME-2 suspension cultures at model-specific optimized densities. After 96 h to allow organoid re-formation, assay plates were prepared with a 50% BME-2:organoid medium base layer, and organoids were transferred into 384-well plates using Multidrop Combi (Thermo Scientific) dispensers. Twenty-four hours later, compounds were dispensed using an Echo555 (Labcyte), and cells were treated for 72 h. Viability was measured using CellTiter-Glo 2.0 (Promega).

Two independent high-throughput screening projects were conducted. The first project used a full 7 × 7 concentration matrix (49 measurements) for each drug combination, and the second used a reduced 25 measurement concentration matrix encompassing the same concentration range. Compounds were tested in biological duplicates in the first project and single replicates in the second project, with higher technical replication for agents included in multiple combinations like afatinib (also tested as monotherapies).

Raw viability data were analysed independently for each project. Data were normalized per plate using negative (untreated, DMSO) and positive (MG-132, staurosporine, blank) controls. Dose-response curves were fitted using a non-linear mixed-effects model to estimate IC₅₀ and AUC values using the gdscIC50 R package (v1.7.3)¹⁰². For combination treatments, the maximum combination effect (combo_MaxE) was defined as the second-highest measured inhibition. Bliss excess was calculated as the difference between observed combination inhibition and the predicted Bliss additivity of the corresponding monotherapies. The results presented are the mean values across both projects.

Validation drug sensitivity testing

For validation drug sensitivity testing, organoids were dissociated into single cells, resuspended in organoid medium, and seeded into 96-well plates over a 50% BME-2:organoid medium base layer (2,000–5,000 cells per well, depending on the model). For monotherapy treatments, nine drug concentrations spanning a 256-fold range were added four days after plating, in technical triplicates. For combination treatments, five concentrations of KRAS inhibitors (sotorasib or MRTX1133) across a 256-fold range were tested with two fixed concentrations of EGFR and ERBB2 inhibitors (afatinib and gefitinib), also in triplicate. Viability was quantified using CellTiter-Glo 2.0 at 72 h post-treatment for monotherapies and at 0, 3, 6 and 9 days for combinations. Dose-response curves were fitted using GraphPad Prism, with three technical and two biological replicates per condition.

EGF depletion from the medium

Organoids were dissociated into single cells, resuspended in organoid culture medium either supplemented with or deprived of EGF, and seeded into 96-well plates over a 50:50 medium-to-ECM layer. The bottom ECM-containing layer was prepared with the same EGF condition as the overlaid medium. Cell viability was quantified using CellTiter-Glo 2.0 seven days after seeding.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

1. Tuveson, D. & Clevers, H. Cancer modeling meets human organoid technology. Science 364, 952–955 (2019).

   Article  ADS  PubMed  CAS  Google Scholar 

2. Dharia, N. V. et al. A first-generation pediatric cancer dependency map. Nat. Genet. 53, 529–538 (2021).

   Article  PubMed  PubMed Central  CAS  Google Scholar 

3. Behan, F. M. et al. Prioritization of cancer therapeutic targets using CRISPR–Cas9 screens. Nature 568, 511–516 (2019).

   Article  ADS  PubMed  CAS  Google Scholar 

4. Ghandi, M. et al. Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 569, 503–508 (2019).

   Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

5. Iorio, F. et al. A landscape of pharmacogenomic interactions in cancer. Cell 166, 740–754 (2016).

   Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

6. Jaaks, P. et al. Effective drug combinations in breast, colon and pancreatic cancer cells. Nature 603, 166–173 (2022).

   Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

7. Corsello, S. M. et al. Discovering the anti-cancer potential of non-oncology drugs by systematic viability profiling. Nat Cancer 1, 235–248 (2020).

   Article  PubMed  PubMed Central  CAS  Google Scholar 

8. Tsherniak, A. et al. Defining a cancer dependency map. Cell 170, 564–576 (2017).

   Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

9. Pacini, C. et al. A comprehensive clinically informed map of dependencies in cancer cells and framework for target prioritization. Cancer Cell 42, 301–316 (2024).

   Article  PubMed  CAS  Google Scholar 

10. Xie, X., Li, X. & Song, W. Tumor organoid biobank-new platform for medical research. Sci Rep. 13, 1819 (2023).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

11. Geevimaan, K. et al. Patient-derived organoid serves as a platform for personalized chemotherapy in advanced colorectal cancer patients. Front. Oncol. 12, 883437 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

12. Yao, Y. et al. Patient-derived organoids predict chemoradiation responses of locally advanced rectal cancer. Cell Stem Cell 26, 17–26 (2020).

    Article  PubMed  CAS  Google Scholar 

13. Larsen, B. M. et al. A pan-cancer organoid platform for precision medicine. Cell Rep. 36, 109429 (2021).

    Article  PubMed  CAS  Google Scholar 

14. Al Shihabi, A. et al. The landscape of drug sensitivity and resistance in sarcoma. Cell Stem Cell 31, 1524–1542 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

15. Milne, J. V., Mustafa, E. H. & Clemons, N. J. Modelling esophageal adenocarcinoma and Barrett’s esophagus with patient-derived organoids. Front. Mol. Biosci. 11, 1382070 (2024).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

16. Yee, C., Dickson, K.-A., Muntasir, M. N., Ma, Y. & Marsh, D. J. Three-dimensional modelling of ovarian cancer: from cell lines to organoids for discovery and personalized medicine. Front. Bioeng. Biotechnol. 10, 836984 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

17. ElHarouni, D. et al. A compendium of next-generation patient-derived models for diverse cancers. Nature https://doi.org/10.1038/s41586-026-10806-y (2026)

18. White, A. et al. A review of sex-related differences in colorectal cancer incidence, screening uptake, routes to diagnosis, cancer stage and survival in the UK. BMC Cancer 18, 906 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

19. Mathieu, L. N., Kanarek, N. F., Tsai, H.-L., Rudin, C. M. & Brock, M. V. Age and sex differences in the incidence of esophageal adenocarcinoma: results from the Surveillance, Epidemiology, and End Results (SEER) Registry (1973–2008). Dis. Esophagus 27, 757–763 (2014).

    Article  PubMed  CAS  Google Scholar 

20. Fan, Y., Yuan, J.-M., Wang, R., Gao, Y.-T. & Yu, M. C. Alcohol, tobacco, and diet in relation to esophageal cancer: the Shanghai Cohort Study. Nutr. Cancer 60, 354–363 (2008).

    Article  PubMed  PubMed Central  Google Scholar 

21. Sinicrope, F. A. Increasing incidence of early-onset colorectal cancer. N. Engl. J. Med. 386, 1547–1558 (2022).

    Article  PubMed  CAS  Google Scholar 

22. Riihimäki, M., Hemminki, A., Sundquist, J. & Hemminki, K. Patterns of metastasis in colon and rectal cancer. Sci Rep. 6, 29765 (2016).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

23. Boland, C. R. & Goel, A. Microsatellite instability in colorectal cancer. Gastroenterology 138, 2073–2087 (2010).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

24. Laughney, A. M., Elizalde, S., Genovese, G. & Bakhoum, S. F. Dynamics of tumor heterogeneity derived from clonal karyotypic evolution. Cell Rep. 12, 809–820 (2015).

    Article  PubMed  CAS  Google Scholar 

25. ICGC/TCGA Pan-Cancer Analysis of Whole Genomes Consortium Pan-cancer analysis of whole genomes. Nature 578, 82–93 (2020).

    Article  ADS  Google Scholar 

26. Cortés-Ciriano, I. et al. Comprehensive analysis of chromothripsis in 2,658 human cancers using whole-genome sequencing. Nat. Genet. 52, 331–341 (2020).

    Article  PubMed  PubMed Central  Google Scholar 

27. Fearon, E. R. & Vogelstein, B. A genetic model for colorectal tumorigenesis. Cell 61, 759–767 (1990).

    Article  PubMed  CAS  Google Scholar 

28. Drost, J. et al. Sequential cancer mutations in cultured human intestinal stem cells. Nature 521, 43–47 (2015).

    Article  ADS  PubMed  CAS  Google Scholar 

29. Lee, J. K. et al. Comprehensive pan-cancer genomic landscape of KRAS altered cancers and real-world outcomes in solid tumors. NPJ Precis. Oncol. 6, 91 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

30. Díaz-Gay, M. et al. Assigning mutational signatures to individual samples and individual somatic mutations with SigProfilerAssignment. Bioinformatics 39, btad756 (2025).

    Article  Google Scholar 

31. Pleguezuelos-Manzano, C. et al. Mutational signature in colorectal cancer caused by genotoxic pks⁺ E. coli. Nature 580, 269–273 (2020).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

32. Dziubańska-Kusibab, P. J. et al. Colibactin DNA-damage signature indicates mutational impact in colorectal cancer. Nat. Med. 26, 1063–1069 (2020).

    Article  PubMed  Google Scholar 

33. Cornish, A. J. et al. The genomic landscape of 2,023 colorectal cancers. Nature 633, 127–136 (2024).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

34. Nguyen, L., Martens, W. J., Van Hoeck, A. & Cuppen, E. Pan-cancer landscape of homologous recombination deficiency. Nat. Commun. 11, 5584 (2020).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

35. Levatić, J., Salvadores, M., Fuster-Tormo, F. & Supek, F. Mutational signatures are markers of drug sensitivity of cancer cells. Nat. Commun. 13, 2926 (2022).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

36. Guinney, J. et al. The consensus molecular subtypes of colorectal cancer. Nat. Med. 21, 1350–1356 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

37. Isella, C. et al. Selective analysis of cancer-cell intrinsic transcriptional traits defines novel clinically relevant subtypes of colorectal cancer. Nat. Commun. 8, 15107 (2017).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

38. Price, S. et al. A suspension technique for efficient large-scale cancer organoid culturing and perturbation screens. Sci Rep. 12, 5571 (2022).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

39. Li, X. et al. Organoid cultures recapitulate esophageal adenocarcinoma heterogeneity providing a model for clonality studies and precision therapeutics. Nat. Commun. 9, 2983 (2018).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

40. van der Meer, D. et al. Cell Model Passports—a hub for clinical, genetic and functional datasets of preclinical cancer models. Nucleic Acids Res. 47, D923–D929 (2019).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

41. Gonçalves, E. et al. Minimal genome-wide human CRISPR–Cas9 library. Genome Biol. 22, 40 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

42. Ringel, T. et al. Genome-scale CRISPR screening in human intestinal organoids identifies drivers of TGF-β resistance. Cell Stem Cell 26, 431–440 (2020).

    Article  PubMed  CAS  Google Scholar 

43. Michels, B. E. et al. Pooled in vitro and in vivo CRISPR–Cas9 screening identifies tumor suppressors in human colon organoids. Cell Stem Cell 26, 782–792 (2020).

    Article  PubMed  CAS  Google Scholar 

44. Geurts, M. H. & Clevers, H. CRISPR engineering in organoids for gene repair and disease modelling. Nat. Rev. Bioeng. 1, 32–45 (2023).

    Article  CAS  Google Scholar 

45. Kim, E. & Hart, T. Improved analysis of CRISPR fitness screens and reduced off-target effects with the BAGEL2 gene essentiality classifier. Genome Med. 13, 2 (2021).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

46. Wang, B. et al. Phospholipid remodeling and cholesterol availability regulate intestinal stemness and tumorigenesis. Cell Stem Cell 22, 206–220 (2018).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

47. Lek, M. et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature 536, 285–291 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

48. Karczewski, K. J. et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature 581, 434–443 (2020).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

49. Picco, G. et al. Novel WRN helicase inhibitors selectively target microsatellite-unstable cancer cells. Cancer Discov. 14, 1457–1475 (2024).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

50. Rustgi, N. et al. Molecular landscape and clinical implication of CCNE1-amplified esophagogastric cancer. Cancer Res. Commun. 4, 1399–1409 (2024).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

51. Chan, P. Y. et al. The synthetic lethal interaction between CDS1 and CDS2 is a vulnerability in uveal melanoma and across multiple tumor types. Nat. Genet. 57, 1672–1683 (2025).

    Article  PubMed  PubMed Central  Google Scholar 

52. Singhal, A., Li, B. T. & O’Reilly, E. M. Targeting KRAS in cancer. Nat. Med. 30, 969–983 (2024).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

53. Marasco, M. et al. Direct inhibition of RAS reveals the features of oncogenic signaling driven by RAS G12 and Q61 mutations. Cancer Discov. 15, 1392–1409 (2025).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

54. Dijkstra, K. K. et al. Generation of tumor-reactive T cells by co-culture of peripheral blood lymphocytes and tumor organoids. Cell 174, 1586–1598 (2018).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

55. Moorman, A. et al. Progressive plasticity during colorectal cancer metastasis. Nature 637, 947–954 (2025).

    Article  ADS  PubMed  CAS  Google Scholar 

56. Lo, Y.-H. et al. Large-scale CRISPR screening in primary human 3D gastric organoids enables comprehensive dissection of gene-drug interactions. Nat. Commun. 16, 7566 (2025).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

57. Hunter, J. C. et al. Biochemical and structural analysis of common cancer-associated KRAS mutations. Mol Cancer Res 13, 1325–1335 (2015).

    Article  PubMed  CAS  Google Scholar 

58. Cancer Genome Atlas Research Network et al The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 45, 1113–1120 (2013).

    Article  Google Scholar 

59. International Cancer Genome Consortium et al International network of cancer genome projects. Nature 464, 993–998 (2010).

    Article  ADS  Google Scholar 

60. Roohani, Y. H. et al. Virtual Cell Challenge: toward a Turing test for the virtual cell. Cell 188, 3370–3374 (2025).

    Article  PubMed  CAS  Google Scholar 

61. Bunne, C. et al. How to build the virtual cell with artificial intelligence: Priorities and opportunities. Cell 187, 7045–7063 (2024).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

62. Rogers, H. et al. Derivation of organoids from primary tumour tissue v1. protocols.io https://doi.org/10.17504/protocols.io.bfvnjn5e (2020).

63. Rogers, H. et al. Derivation of organoids from frozen tumour material v1. protocols.io https://doi.org/10.17504/protocols.io.biiakcae (2020).

64. Rogers, H. et al. Passaging cancer organoid cultures v1. protocols.io https://doi.org/10.17504/protocols.io.bfe3jjgn (2020).

65. van de Wetering, M. et al. Prospective derivation of a living organoid biobank of colorectal cancer patients. Cell 161, 933–945 (2015).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

66. Boj, S. F. et al. Organoid models of human and mouse ductal pancreatic cancer. Cell 160, 324–338 (2015).

    Article  PubMed  CAS  Google Scholar 

67. Trillsch, F. et al. Protocol to optimize the biobanking of ovarian cancer organoids by accommodating patient-specific differences in stemness potential. STAR Protoc. 4, 102484 (2023).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

68. Li, H. & Durbin, R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 26, 589–595 (2010).

    Article  PubMed  PubMed Central  Google Scholar 

69. Jones, D. et al. cgpCaVEManWrapper: simple execution of CaVEMan in order to detect somatic single nucleotide variants in NGS data. Curr. Protoc. Bioinformatics 56, 15.10.1–15.10.18 (2016).

    PubMed  PubMed Central  Google Scholar 

70. Raine, K. M. et al. cgpPindel: identifying somatically acquired insertion and deletion events from paired end sequencing. Curr. Protoc. Bioinformatics 52, 15.7.1–15.7.12 (2015).

    PubMed  PubMed Central  Google Scholar 

71. Yates, L. R. et al. Genomic evolution of breast cancer metastasis and relapse. Cancer Cell 32, 169–184 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

72. Priestley, P. et al. Pan-cancer whole-genome analyses of metastatic solid tumours. Nature 575, 210–216 (2019).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

73. Cingolani, P. et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118; iso-2; iso-3. Fly 6, 80–92 (2012).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

74. Cameron, D. L. et al. GRIDSS2: comprehensive characterisation of somatic structural variation using single breakend variants and structural variant phasing. Genome Biol. 22, 202 (2021).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

75. Tarailo-Graovac, M. & Chen, N. Using RepeatMasker to identify repetitive elements in genomic sequences. Curr. Protoc. Bioinformatics 4, 4.10.1–4.10.14 (2009).

    Google Scholar 

76. Lu, J. et al. Metagenome analysis using the Kraken software suite. Nat. Protoc. 17, 2815–2839 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

77. Shale, C. et al. Unscrambling cancer genomes via integrated analysis of structural variation and copy number. Cell Genomics 2, 100112 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

78. Muiños, F., Martínez-Jiménez, F., Pich, O., Gonzalez-Perez, A. & Lopez-Bigas, N. In silico saturation mutagenesis of cancer genes. Nature 596, 428–432 (2021).

    Article  ADS  PubMed  Google Scholar 

79. Tate, J. G. et al. COSMIC: the Catalogue Of Somatic Mutations In Cancer. Nucleic Acids Res. 47, D941–D947 (2019).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

80. Chang, M. T. et al. Accelerating discovery of functional mutant alleles in cancer. Cancer Discov. 8, 174–183 (2018).

    Article  PubMed  CAS  Google Scholar 

81. Huang, K.-L. et al. Pathogenic germline variants in 10,389 adult cancers. Cell 173, 355–370 (2018).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

82. Steele, C. D. et al. Undifferentiated sarcomas develop through distinct evolutionary pathways. Cancer Cell 35, 441–456 (2019).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

83. Dentro, S. C. et al. Characterizing genetic intra-tumor heterogeneity across 2,658 human cancer genomes. Cell 184, 2239–2254 (2021).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

84. Amemiya, H. M., Kundaje, A. & Boyle, A. P. The ENCODE Blacklist: identification of problematic regions of the genome. Sci Rep. 9, 9354 (2019).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

85. Islam, S. M. A. et al. Uncovering novel mutational signatures by de novo extraction with SigProfilerExtractor. Cell Genomics 2, 100179 (2022).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

86. Díaz-Gay, M. et al. Geographic and age variations in mutational processes in colorectal cancer. Nature 643, 230–240 (2025).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

87. Cagan, A. et al. Somatic mutation rates scale with lifespan across mammals. Nature 604, 517–524 (2022).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

88. Alexandrov, L. B. et al. The repertoire of mutational signatures in human cancer. Nature 578, 94–101 (2020).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

89. Espejo Valle-Inclan, J. et al. Mechanisms underpinning osteosarcoma genome complexity and evolution. Preprint at bioRxiv https://doi.org/10.1101/2023.12.29.573403 (2023).

90. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).

    Article  PubMed  CAS  Google Scholar 

91. Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics 12, 323 (2011).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

92. Eide, P. W., Bruun, J., Lothe, R. A. & Sveen, A. CMScaller: an R package for consensus molecular subtyping of colorectal cancer pre-clinical models. Sci Rep. 7, 16618 (2017).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

93. Fowler, T., Smith, J., Jackson, A., Souster, E. & Rogers, H. Whole-genome CRISPR screening of stably expressing Cas9 cancer organoid lines v2. protocols.io https://doi.org/10.17504/protocols.io.3byl4q6zrvo5/v2 (2023).

94. Smith, J., Fowler, T., Andres, A., Souster, E. & Rogers, H. Generating stably-expressing Cas9 cancer organoid lines v2. protocols.io https://doi.org/10.17504/protocols.io.6qpvr3ydovmk/v2 (2023).

95. Tzelepis, K. et al. A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets in acute myeloid leukemia. Cell Rep. 17, 1193–1205 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

96. Iorio, F. et al. Unsupervised correction of gene-independent cell responses to CRISPR–Cas9 targeting. BMC Genomics 19, 604 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

97. Johnson, W. E., Li, C. & Rabinovic, A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8, 118–127 (2007).

    Article  PubMed  Google Scholar 

98. Vinceti, A. et al. CoRe: a robustly benchmarked R package for identifying core-fitness genes in genome-wide pooled CRISPR–Cas9 screens. BMC Genomics 22, 828 (2021).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

99. Kanehisa, M., Furumichi, M., Sato, Y., Kawashima, M. & Ishiguro-Watanabe, M. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 51, D587–D592 (2023).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

100. Liberzon, A. et al. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1, 417–425 (2015).

     Article  PubMed  PubMed Central  CAS  Google Scholar 

101. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).

     Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

102. Vis, D. J. et al. Multilevel models improve precision and speed of IC₅₀ estimates. Pharmacogenomics 17, 691–700 (2016).

     Article  PubMed  PubMed Central  CAS  Google Scholar 

103. Herranz-Ors, C, Bhosle, S. & Garnett, M. A tumour-derived organoid biobank maps cancer gene dependencies. Figshare https://doi.org/10.6084/m9.figshare.28339340 (2026).

Download references

Author notes

1. L. Letchford & S. F. Vieira

   Present address: AstraZeneca, The Discovery Centre (DISC), Cambridge, UK

2. H. E. Francies

   Present address: GlaxoSmithKline (GSK), Stevenage, UK

Authors and Affiliations

1. Wellcome Sanger Institute, Cambridge, UK

   C. Herranz-Ors, S. G. Bhosle, A. E. Beck, J. G. R. Gilbert, G. Picco, S. Valentini, A. E. Andres, R. Ansari, S. Barthorpe, G. Battarbee, C. M. Beaver, S. Brocklesby, J. Cantwell, C. A. Collins, J. Davis, H. G. Dimitrova, J. Doran, E. Efendi, K. Evans, M. Fekry, T. A. Fowler, M. Garcia-Casado, J. A. T. Griffiths, C. Hall, R. Hamer, C. Hardy, Z. Hewitson, E. Hitch, L. Holland, D. A. Jackson, N. Joshi, A. Kavasakali, L. Letchford, H. B. Lightfoot, H. Lingala, I. Mali, K. May, T. Mironenko, J. Morris, C. Pacini, S. Price, G. Robert-Tissot, H. A. Rogers, J. V. Smith, K. Smith, E. Souster, W. J. Spence, F. Thomas, S. F. Vieira, S. Walker, G. Alfonsin, M. R. Stratton, U. McDermott, I. Cortes-Ciriano, H. E. Francies & M. J. Garnett

2. European Molecular Biology Laboratory, European Bioinformatics Institute, Cambridge, UK

   J. Espejo Valle-Inclan, F. Muyas & I. Cortes-Ciriano

3. Cancer Research UK Cambridge Institute, University of Cambridge, Li Ka Shing Centre, Cambridge, UK

   J. A. T. Griffiths, J. Hall, D. A. Sanders, M. Vias, P. A. W. Edwards, M. Eldridge, M. Secrier, G. Devonshire, S. Jammula & J. D. Brenton

4. AstraZeneca, The Discovery Centre (DISC), Cambridge, UK

   E. Souster & U. McDermott

5. Instituto de Investigación Biomédica A Coruña (INIBIC), A Coruña, Spain

   G. Alfonsin

6. University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK

   H. Bermingham, E. A. Griffiths, E. Y. L. Leung, C. Millington, K. Roberts, P. Taniere, K. Wanigasooriya, O. Tucker, A. Beggs, S. Puig, G. Contino & A. D. Beggs

7. Early Cancer Institute, University of Cambridge, Cambridge, UK

   H. Coles, A. Freeman, N. Grehan, C. Loreno, B. Nutzinger, A. M. Redmond, P. A. W. Edwards, S. Abbas, M. O’Donovan, A. Miremadi, S. Malhotra, M. Tripathi, C. Cheah & R. C. Fitzgerald

8. Department of Surgery and Cancer, Imperial College London, London, UK

   D. P. Ennis, G. Giannone, G. B. Hanna, C. J. Peters, K. Moorthy & I. A. McNeish

9. Institute of Immunology and Immunotherapy, College of Medical and Dental Science, University of Birmingham, Birmingham, UK

   E. A. Griffiths & K. Roberts

10. School of Cancer Sciences, Faculty of Medicine, University of Southampton, Southampton, UK

    S. L. Lee, A. Mirnezami, S. Roy & T. J. Underwood

11. Department of Cancer and Genomic Sciences, College of Medicine and Health, University of Birmingham, Birmingham, UK

    E. Y. L. Leung, K. Orzechowska, C. M. A. Pinna, A. Beggs & A. D. Beggs

12. University Hospital Southampton NHS Foundation Trust, Southampton, UK

    S. Roy & B. L. Grace

13. George Eliot Hospital NHS Trust, Nuneaton, UK

    K. Wanigasooriya

14. Lymphoid Malignancies Branch, Center for Cancer Research, National Cancer Institute, Bethesda, MD, USA

    L. M. Staudt

15. School of Cancer Sciences, University of Glasgow, Glasgow, UK

    A. Biankin & O. J. Sansom

16. Cancer Research UK Scotland Institute, Glasgow, UK

    O. J. Sansom

17. Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK

    E. C. Smyth, N. Carroll, R. H. Hardwick, P. Safranek, A. Hindmarsh, V. Sujendran & J. R. O’Neill

18. Department of Histopathology, Addenbrookes Hospital, Cambridge, UK

    M. O’Donovan, A. Miremadi, S. Malhotra, M. Tripathi, B. Kumar & L. Sreedharan

19. Department of Computer Science, University of Oxford, Oxford, UK

    J. Davies, C. Crichton, S. L. Parsons, I. Soomro & P. Kaye

20. Salford Royal NHS Foundation Trust, Salford, UK

    S. J. Hayes, Y. Ang & J. Saunders

21. Faculty of Medical and Human Sciences, University of Manchester, Manchester, UK

    S. J. Hayes

22. Wigan and Leigh NHS Foundation Trust, Wigan, UK

    Y. Ang

23. GI Science Centre, University of Manchester, Manchester, UK

    Y. Ang & A. Sharrocks

24. Royal Surrey County Hospital NHS Foundation Trust, Guildford, UK

    S. R. Preston & I. Bagwan

25. Edinburgh Royal Infirmary, Edinburgh, UK

    V. Save, R. J. E. Skipworth & J. R. O’Neill

26. Edinburgh University, Edinburgh, UK

    J. R. O’Neill

27. Heart of England NHS Foundation Trust, Birmingham, UK

    O. Tucker

28. Guy’s and St Thomas’s NHS Foundation Trust, London, UK

    J. Lagergren, J. Gossage, A. Davies, F. Chang & U. Mahadeva

29. Karolinska Institute, Stockholm, Sweden

    J. Lagergren

30. King’s College London, London, UK

    J. Gossage, A. Davies, F. Chang, V. Goh & F. D. Ciccarelli

31. Plymouth Hospitals NHS Trust, Plymouth, UK

    G. Sanders & D. Chan

32. Norfolk and Norwich University Hospital NHS Foundation Trust, Norwich, UK

    E. Cheong

33. Nottingham University Hospitals NHS Trust, Nottingham, UK

    J. Saunders

34. University College London, London, UK

    L. Lovat & R. Haidry

35. Wythenshawe Hospital, Manchester, UK

    M. Scott

36. University Hospitals Coventry and Warwickshire NHS Trust, Coventry, UK

    S. Sothi

37. Queen’s Medical Centre, University of Nottingham, Nottingham, UK

    A. Grabowska

38. Centre for Cancer Research and Cell Biology, Queen’s University Belfast, Belfast, UK

    R. Turkington, D. McManus & H. Coleman

39. Tayside Cancer Centre, Ninewells Hospital and Medical School, Dundee, Scotland

    R. D. Petty

40. Portsmouth Hospitals NHS Trust, Portsmouth, UK

    F. Bartlett

41. Velindre University NHS Trust, Cardiff, UK

    T. D. L. Crosby

Consortia

Contributions

Conceptualization: C.H.-O., C.M.A.P., M.V., M.R.S., L.M.S, U.M., J.D.B., A.B., O.J.S., T.J.U., R.C.F., A.D.B., H.E.F. and M.J.G. Data curation: C.H.-O., S.G.B., J.G.R.G., J.E.V.-I., F.M., S. Barthorpe, H.B.L., C.P. and I.C.-C. Formal analysis: C.H.-O., S.G.B., J.G.R.G., J.E.V.-I., F.M. and I.C.-C. Funding acquisition: G.G., M.R.S., U.M., I.A.M., I.C.-C., H.E.F. and M.J.G. Organoid derivation: C.A.C., J. Davis, E.E., M.F., M.G.-C., J.A.T.G., C. Hall, L.H., N.J., L.L., H.A.R., W.J.S. and S.F.V. Drug screening: A.E.B., G.B., S. Brocklesby, J.C., H.G.D., K.E., M.G.-C., C. Hall, E.H., A.K., I.M., K.M., T.M., J.M., G.R.-T., F.T. and S.W. CRISPR screening: A.E.A., R.A., T.A.F., J.A.T.G., D.A.J., J.V.S., E.S. and S.F.V. Pellet extraction and PCR: K.S. Sample and data acquisition: H.C., D.P.E., A.F., J.A.T.G., G.G., N.G., S.L.L., C.L., C.M.A.P., C.M., A.M., B.N., S.R., I.A.M. and T.J.U. Computational methodology: S.G.B., J.G.R.G., H.B.L. and C.P. Experimental methodology: A.E.A., R.A., S. Barthorpe, C.A.C., J. Davis, M.G.-C., L.H., D.A.J., L.L., S.P., H.A.R., S.F.V., H.B., D.P.E., E.A.G., J.H., E.Y.L.L., K.O., C.M.A.P., K.R., D.A.S., P.T., M.V. and K.W. Project administration: A.E.B., S.V., S. Barthorpe, C.M.B., J. Doran, R.H., C. Hardy, Z.H., H.L., T.M., H.A.R., D.P.E., C.M.A.P., A.M.R., S.R., I.A.M., I.C.-C., H.E.F. and M.J.G. Software: C.H.-O., S.G.B., J.G.R.G., J.E.V.-I., F.M. and I.C.-C. Supervision: A.E.B., C.M.A.P., M.V., G.P., J.D.B., I.A.M., A.B., O.J.S., I.C.-C., T.J.U., R.C.F., A.D.B., H.E.F. and M.J.G. Validation: C.H.-O., G.A., J.E.V.-I., F.M., G.P. and I.C.-C. Visualization: C.H.-O., J.E.V.-I., F.M., T.A.F., H.A.R., J.V.S. and I.C.-C. Writing, original draft: C.H.-O. and M.J.G. Writing, review and editing: C.H.-O., A.E.B., S.G.B., C.P., G.P., D.P.E., G.G., I.A.M., I.C.-C., R.C.F., A.D.B., H.E.F. and M.J.G.

Corresponding author

Correspondence to M. J. Garnett.

Competing interests

M.J.G. is a consultant and equity holder for Mosaic Therapeutics and is an inventor on a patent titled Culture of Organoids (patent ID GB2596787) that has been licensed by Genome Research Limited to Mosaic Therapeutics. R.C.F. is named on patents related to Cytosponge and related assays that have been licensed by the Medical Research Council to Covidien GI Solutions (now Medtronic) and is a co-founder and shareholder (<3%) of CYTED Ltd. R.C.F. has an ongoing collaboration with AstraZeneca and Natera with provision of expertise and materials. R.C.F. is a member of the scientific advisory board for the AstraZeneca and CRUK Functional Genomics Core and is on an advisory board for 23andme. J.D.B is a founder and shareholder of Tailor Bio Ltd.

Peer review information

Nature thanks Andreas Moor, Toshiro Sato, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions