Pharmacogenomic Datasets Research Articles

Abstract 7365: Drug-centric prior improves drug response signature identification in partially overlapping, large-scale pharmacogenomic datasets

Abstract Introduction: With the accumulation of large-scale pharmacogenomic data such as whole-genome RNA sequencing, copy number and mutation profiles for tens of thousands of samples, further screened with thousands of small molecules and other perturbagens, the question arises how to best leverage partially overlapping datasets generated at different facilities. Notably, Haibe-Kains et al. observed discordance in state-of-the-art pharmacogenomic repositories. As research groups across the world continue to generate drug screens of variable size and quality, the need for approaches that can learn from such partially overlapping experiments and improve the signal to noise ratio emerges with increasing importance. Methods: A previously published Bayesian group factor analysis model was shown to outperform other approaches in predicting drug response and identifying gene signature from pharmacogenomic datasets particularly by leveraging shared information for a given gene across multiple omics assays. Here, we applied the same model in a similar fashion to now learn from shared observations for the same drug across multiple partially overlapping and noisy small molecule screens. We integrate gene expression, mutation and drug response data from the two largest pan-cancer repositories at the Broad and Sanger Institutes, respectively. We train joint models on partially overlapping data from both. We evaluate our performance in three ways: 1) we test if the joint model improves the prioritization of known consensus biomarkers for the drugs shared between the two cohorts; 2) we test if the joint model improves the recapitulation of shared drug mechanism of action as compared to the single dataset models; 3) we evaluate the joint models with respect to pathway enrichment as compared to the single dataset models. Results: We evaluated the performance of our joint model for 5 drugs shared between the 2 resources: selumetinib, tanespimycin, nutlin 3A, mirdametinib and PLX4720. First, we show that training joint models on partially overlapping pharmacogenomic datasets can overall improve gene signature identification by improving the ranks of known consensus biomarkers. Second, we show that the joint model learns a latent representation of the drugs that better recapitulates the underlying known mechanisms of action for the three serine/threonine kinase inhibitors. Finally, we show that the joint model achieved improved pathway enrichment results for the targeted MAPK/ERK signaling pathway. Conclusions: We present an application of a Bayesian group factor analysis model, where we employ a drug-centric prior to transfer information about drugs screened in multiple datasets. We show that joint models leveraging partially overlapping large-scale pharmacogenomic datasets from the Broad and Sanger institutes can overall improve drug signature identification. Citation Format: Dharani Thirumalaisamy, Sunil K. Joshi, Mehmet Gönen, Olga Nikolova. Drug-centric prior improves drug response signature identification in partially overlapping, large-scale pharmacogenomic datasets [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7365.

Abstract 7175: Web-based cross-database exploration of molecular pharmacology data from cancer cell lines and patient genomics

Abstract CellMiner Cross-Database (CellMinerCDB, discover.nci.nih.gov/cellminercdb) allows integration and analysis of molecular and pharmacological data within and across cancer cell line datasets from the National Cancer Institute (NCI), Broad Institute, Sanger/MGH, MD Anderson Cancer Center (MDACC), and the National Center for Advancing Translational Sciences (NCATS, discover.nci.nih.gov/rsconnect/cellminercdb_ncats/). Altogether, our database spans over 1,700 distinct cell lines, over 1,000 clinically relevant drugs, ~25,000 experimental compounds, and molecular profiling data, such as gene/protein expression, molecular signatures including epithelial mesenchymal transition (EMT), replication stress (RepStress), neuroendocrine evaluation (NE), DNA copy, methylation, mutational status, and gene dependencies (Broad DepMap). This rich and growing set of cell lines and tested drugs allows for novel avenues for response determinant discovery and clinical translation that is facilitated by 1) common annotations (e.g., tissues of origin and matched drug names) across pharmacogenomic datasets and 2) various univariate and multivariate analyses tools. A major challenge to clinical translation occurs in relating pre-clinical models to patient samples. As part of this work, we show how the CellMinerCDB infrastructure additionally supports annotated patient multi-omics data to aid researchers to relate findings from cancer cell lines with patient data; we use data from small cell lung cancer (SCLC) and neuroendocrine tumors as illustrative examples. This work provides significant pharmacogenomic integration that allows exploration within a comprehensive dataset of cancer cell lines and to inter-relationships with patient-derived datasets; this is being developed to display and mine patient samples as part of the complementary web-based tool (PatientTumorMiner). Citation Format: William C. Reinhold, Augustin Luna, Fathi Elloumi, Sudhir Varma, Daiki Taniyama, Makito Mizunuma, Jaydira del Rivero, Anish Thomas, Karyyne Reilly, Karel Pacak, Mirit Aladjem, Yves Pommier. Web-based cross-database exploration of molecular pharmacology data from cancer cell lines and patient genomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7175.

Abstract 907: ScreenDL: A transfer learning framework integrating tumor omics and functional drug screening for personalized clinical drug response prediction

Abstract Precision oncology hinges on accurate prediction of patient-specific treatment response from tumor molecular inputs. While deep learning (DL) models achieve state-of-the-art response prediction in cell lines, existing methods do not readily translate to the clinic where training data is limited. Here, we have developed and validated ScreenDL, a novel DL framework designed explicitly for use in clinical precision oncology applications. The underlying architecture of ScreenDL consists of two fully connected branches dedicated to learning rich embeddings of a drug’s chemical structure and a tumor’s transcriptomic profile respectively. These tumor omic and drug molecular embeddings are fed into a shared response prediction subnetwork which outputs predicted ln(IC50) values. The ScreenDL training schema incorporates three phases, each designed to improve clinical response prediction: (1) an initial pretraining phase leverages large-scale cell line pharmacogenomic datasets to extract patterns/relationships linking tumor omics and drug response; (2) subsequent transfer learning integrates pharmacogenomic data from patient-derived models of cancer, adapting the pretrained model to a more patient-relevant context; (3) patient-level omics and drug screening data (e.g., from matched patient-derived organoid (PDO) models) is integrated through patient-specific fine-tuning, generating a patient-specific response prediction model. Importantly, biomaterial from surgical tumor resection is often available for organoid establishment and patient-specific drug screening in practical clinical scenarios. To assess the utility of our ScreenDL framework for clinical response prediction, we applied ScreenDL to predict treatment response in 50 advanced/metastatic triple negative breast cancer (TNBC) patient-derived xenograft models (PDX). We leveraged tumor omics profiles for each PDX and drug screening data from matched PDX-derived organoids (PDxOs), mirroring the combination of patient-level tumor omic characterization and functional drug screening in matched PDOs, a protocol currently being piloted in Functional Precision Oncology trials at the University of Utah Huntsman Cancer Institute. After cell line pretraining, ScreenDL achieves a modest median Pearson correlation per drug of 0.15 relative to 0.03 for the leading compared model. However, transfer learning and subsequent PDX-specific fine-tuning significantly improve performance, producing median Pearson correlations per drug of 0.39 and 0.51, respectively. These results demonstrate the ability of our ScreenDL framework to dramatically improve response prediction in clinically relevant domains, bringing DL-based precision treatment selection closer to clinical application. Citation Format: Casey Sederman, Tonya Di Sera, Yi Qiao, Xiaomeng Huang, Bryan E. Welm, Alana L. Welm, Gabor Marth. ScreenDL: A transfer learning framework integrating tumor omics and functional drug screening for personalized clinical drug response prediction [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 907.

Text-mining-based feature selection for anticancer drug response prediction.

Predicting anticancer treatment response from baseline genomic data is a critical obstacle in personalized medicine. Machine learning methods are commonly used for predicting drug response from gene expression data. In the process of constructing these machine learning models, one of the most significant challenges is identifying appropriate features among a massive number of genes. In this study, we utilize features (genes) extracted using the text-mining of scientific literatures. Using two independent cancer pharmacogenomic datasets, we demonstrate that text-mining-based features outperform traditional feature selection techniques in machine learning tasks. In addition, our analysis reveals that text-mining feature-based machine learning models trained on in vitro data also perform well when predicting the response of in vivo cancer models. Our results demonstrate that text-mining-based feature selection is an easy to implement approach that is suitable for building machine learning models for anticancer drug response prediction. https://github.com/merlab/text_features.

Addressing Ancestry and Sex Bias in Pharmacogenomics.

The association of an individual's genetic makeup with their response to drugs is referred to as pharmacogenomics. By understanding the relationship between genetic variants and drug efficacy or toxicity, we are able to optimize pharmacological therapy according to an individual's genotype. Pharmacogenomics research has historically suffered from bias and underrepresentation of people from certain ancestry groups and of the female sex. These biases can arise from factors such as drugs and indications studied, selection of study participants, and methods used to collect and analyze data. To examine the representation of biogeographical populations in pharmacogenomic data sets, we describe individuals involved in gene-drug response studies from PharmGKB, a leading repository of drug-gene annotations, and showcaseCYP2D6, a gene that metabolizes approximately 25% of all prescribed drugs. We also show how the historical underrepresentation of females in clinical trials has led to significantly more adverse drug reactions in females than in males.

Validation of transcriptome signature reversion for drug repurposing in oncology.

Transcriptome signature reversion (TSR) has been extensively proposed and used to discover new indications for existing drugs (i.e. drug repositioning, drug repurposing) for various cancer types. TSR relies on the assumption that a drug that can revert gene expression changes induced by a disease back to original, i.e. healthy, levels is likely to be therapeutically active in treating the disease. Here, we aimed to validate the concept of TSR using the PRISM repurposing data set, which is-as of writing-the largest pharmacogenomic data set. The predictive utility of the TSR approach as it has currently been used appears to be much lower than previously reported and is completely nullified after the drug gene expression signatures are adjusted for the general anti-proliferative downstream effects of drug-induced decreased cell viability. Therefore, TSR mainly relies on generic anti-proliferative drug effects rather than on targeting cancer pathways specifically upregulated in tumor types.

Multi-Omics Alleviates the Limitations of Panel Sequencing for Cancer Drug Response Prediction.

Comprehensive genomic profiling using cancer gene panels has been shown to improve treatment options for a variety of cancer types. However, genomic aberrations detected via such gene panels do not necessarily serve as strong predictors of drug sensitivity. In this study, using pharmacogenomics datasets of cell lines, patient-derived xenografts, and ex vivo treated fresh tumor specimens, we demonstrate that utilizing the transcriptome on top of gene panel features substantially improves drug response prediction performance in cancer.

Tissue-specific identification of multi-omics features for pan-cancer drug response prediction

SummaryCurrent statistical models for drug response prediction and biomarker identification fall short in leveraging the shared and unique information from various cancer tissues and multi-omics profiles. We developed mix-lasso model that introduces an additional sample group penalty term to capture tissue-specific effects of features on pan-cancer response prediction. The mix-lasso model takes into account both the similarity between drug responses (i.e., multi-task learning), and the heterogeneity between multi-omics data (multi-modal learning). When applied to large-scale pharmacogenomics dataset from Cancer Therapeutics Response Portal, mix-lasso enabled accurate drug response predictions and identification of tissue-specific predictive features in the presence of various degrees of missing data, drug-drug correlations, and high-dimensional and correlated genomic and molecular features that often hinder the use of statistical approaches in drug response modeling. Compared to tree lasso model, mix-lasso identified a smaller number of tissue-specific features, hence making the model more interpretable and stable for drug discovery applications.

Abstract 5685: Recognition of internal mRNA N7-methylguanosine by QKI shuttles transcripts into stress granules and modulates drug resistance

Abstract Introduction: Albeit that over 170 types of chemical modifications have been identified in RNAs, the biological functions of most of those decorations are still elusive. N7-methylguanosine (m7G), routinely occurring at the 5’ cap of mRNA or within tRNA and rRNA, also exists in internal mRNA. However, the “reader” proteins that recognize internal mRNA m7G and regulate the metabolism and fate of target mRNAs have yet to be identified. Here, we aim to identify the reader(s) of internal mRNA m7G and elucidate the biological function. Methods: To identify internal m7G reader(s), we utilized RNA-pull down, mass spectrometry (MS), RNA immunoprecipitation (RIP)-seq, and m7G-seq. To evaluate the interaction between QKI and G3BP1, we used Co-IP, immunofluorescence (IF) and proximity ligation assay (PLA). To investigate the function of QKI in stress granule (SG), we performed polyA & m7G RNA fluorescence in situ hybridization (FISH), triple quadrupole MS (QQQ-MS), and Ribo-seq. To assess the role of QKI in drug resistance, we carried out both in vitro cell survival assays and in vivo xenograft mouse model. Results: Our QQQ-MS results suggest cytoplasmic mRNA, in contrast to nuclei mRNA, are highly enriched with internal m7G modification. Quaking proteins (QKI), especially QKI7, can selectively recognize the internal mRNA m7G decoration in the cytosol of various cell types. Our high-throughput sequence data revealed over 1000 confident m7G-modified and QKI-binding RNA targets with a conserved motif, “GANGAN (N=A/U/G)”. Those transcripts are functionally enriched in the “pathways in cancer” and “Hippo signaling pathway”. More strikingly, internal m7G reader QKI7 directly interacts with the SG core protein G3BP1 and can shuttle a subset of m7G-modified transcripts into SG mRNA pool under oxidative stress condition. Additionally, we identified 314 confident m7G-modified & QKI7-binding & SG-enriched target genes in U2OS cells. Ribo-seq indicated that QKI7 modulates the translation efficiency of a set of m7G-modified transcripts via sequestering and silencing them in SGs. Moreover, in line with the observation that many chemotherapy drugs could trigger the assembly of SGs, QKI7 increases the sensitivity of cancer cells to drug treatment (e.g., doxorubicin) in vitro and in vivo in a m7G-dependent manner. Consistently, by analyzing the hepatocellular carcinoma (HCC) pharmacogenomic landscape study dataset, we found that HCC cell lines with higher QKI expression levels are more sensitive to a variety of chemotherapy or targeted therapeutic drugs. Conclusion: Our study identifies QKI as the first internal mRNA m7G reader, provides insights into the biological function of internal mRNA m7G modification in remodeling stress granules transcriptome, and highlights the therapeutic potential of targeting the QKI7/m7G axis in overcoming drug resistance. Citation Format: Zhicong Zhao, Lei Dong, Li Han, Yangchan Li, Jianhuang Xue, Keren Zhou, Wei Li, Xiaolan Deng, Ying Qing, Brandon Tan, Zhenhua Chen, Chao Shen, Andrew Small, Kitty Wang, Keith Leung, Zheng Zhang, Xi Qin, Qiang Xia, Rui Su, Jianjun Chen. Recognition of internal mRNA N7-methylguanosine by QKI shuttles transcripts into stress granules and modulates drug resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5685.

Abstract LB519: Bimodality of gene expression in cancer patients as interpretable biomarkers for drug sensitivity

Abstract Identifying biomarkers predictive of cancer cells’ response to drug treatment constitutes one of the main challenges in precision oncology. Recent large-scale cancer pharmacogenomic studies have opened new avenues of research to develop predictive biomarkers by profiling thousands of human cancer cell lines at the molecular level and screening them with hundreds of approved drugs and experimental chemical compounds. Many studies have leveraged these data to build predictive models of response using various statistical and machine learning methods. However, a common challenge in these methods is the lack of interpretability as to how they make the predictions, hindering the clinical translation of these models. To alleviate this issue, we used the recent logic modeling approach to develop a new machine learning pipeline that explores the space of bimodally expressed genes in multiple large in vitro pharmacogenomic studies and builds multivariate, nonlinear, yet interpretable logic-based models predictive of drug response. We showcased the performance of the approach in a compendium of the three largest in vitro pharmacogenomic data sets to build robust and interpretable models for 101 drugs that span 17 drug classes with high validation rate in independent datasets. These results along with in vivo validation support a better translation of gene expression biomarkers between model systems using bimodal genes. Citation Format: Benjamin Haibe-Kains, Wail Ba-alawi. Bimodality of gene expression in cancer patients as interpretable biomarkers for drug sensitivity [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB519.

Machine learning approach informs biology of cancer drug response

BackgroundThe mechanism of action for most cancer drugs is not clear. Large-scale pharmacogenomic cancer cell line datasets offer a rich resource to obtain this knowledge. Here, we present an analysis strategy for revealing biological pathways that contribute to drug response using publicly available pharmacogenomic cancer cell line datasets.MethodsWe present a custom machine-learning based approach for identifying biological pathways involved in cancer drug response. We test the utility of our approach with a pan-cancer analysis of ML210, an inhibitor of GPX4, and a melanoma-focused analysis of inhibitors of BRAFV600. We apply our approach to reveal determinants of drug resistance to microtubule inhibitors.ResultsOur method implicated lipid metabolism and Rac1/cytoskeleton signaling in the context of ML210 and BRAF inhibitor response, respectively. These findings are consistent with current knowledge of how these drugs work. For microtubule inhibitors, our approach implicated Notch and Akt signaling as pathways that associated with response.ConclusionsOur results demonstrate the utility of combining informed feature selection and machine learning algorithms in understanding cancer drug response.

Bimodal Gene Expression in Patients with Cancer Provides Interpretable Biomarkers for Drug Sensitivity.

A new machine learning pipeline exploits the bimodality of gene expression to provide a reliable set of candidate predictive biomarkers with a high potential for clinical translatability.

Reassessing pharmacogenomic cell sensitivity with multilevel statistical models.

Pharmacogenomic experiments allow for the systematic testing of drugs, at varying dosage concentrations, to study how genomic markers correlate with cell sensitivity to treatment. The first step in the analysis is to quantify the response of cell lines to variable dosage concentrations of the drugs being tested. The signal to noise in these measurements can be low due to biological and experimental variability. However, the increasing availability of pharmacogenomic studies provides replicated data sets that can be leveraged to gain power. To do this, we formulate a hierarchical mixture model to estimate the drug-specific mixture distributions for estimating cell sensitivity and for assessing drug effect type as either broad or targeted effect. We use this formulation to propose a unified approach that can yield posterior probability of a cell being susceptible to a drug conditional on being a targeted effect or relative effect sizes conditioned on the cell being broad. We demonstrate the usefulness of our approach via case studies. First, we assess pairwise agreements for cell lines/drugs within the intersection of two data sets and confirm the moderate pairwise agreement between many publicly available pharmacogenomic data sets. We then present an analysis that identifies sensitivity to the drug crizotinib for cells harboring EML4-ALK or NPM1-ALK gene fusions, as well as significantly down-regulated cell-matrix pathways associated with crizotinib sensitivity.

PharmacoDB 2.0: improving scalability and transparency of in vitro pharmacogenomics analysis.

Cancer pharmacogenomics studies provide valuable insights into disease progression and associations between genomic features and drug response. PharmacoDB integrates multiple cancer pharmacogenomics datasets profiling approved and investigational drugs across cell lines from diverse tissue types. The web-application enables users to efficiently navigate across datasets, view and compare drug dose–response data for a specific drug-cell line pair. In the new version of PharmacoDB (version 2.0, https://pharmacodb.ca/), we present (i) new datasets such as NCI-60, the Profiling Relative Inhibition Simultaneously in Mixtures (PRISM) dataset, as well as updated data from the Genomics of Drug Sensitivity in Cancer (GDSC) and the Genentech Cell Line Screening Initiative (gCSI); (ii) implementation of FAIR data pipelines using ORCESTRA and PharmacoDI; (iii) enhancements to drug–response analysis such as tissue distribution of dose–response metrics and biomarker analysis; and (iv) improved connectivity to drug and cell line databases in the community. The web interface has been rewritten using a modern technology stack to ensure scalability and standardization to accommodate growing pharmacogenomics datasets. PharmacoDB 2.0 is a valuable tool for mining pharmacogenomics datasets, comparing and assessing drug–response phenotypes of cancer models.

Predicting Cancer Drug Response In Vivo by Learning an Optimal Feature Selection of Tumour Molecular Profiles.

(1) Background: Inter-tumour heterogeneity is one of cancer’s most fundamental features. Patient stratification based on drug response prediction is hence needed for effective anti-cancer therapy. However, single-gene markers of response are rare and/or may fail to achieve a significant impact in the clinic. Machine Learning (ML) is emerging as a particularly promising complementary approach to precision oncology. (2) Methods: Here we leverage comprehensive Patient-Derived Xenograft (PDX) pharmacogenomic data sets with dimensionality-reducing ML algorithms with this purpose. (3) Results: Combining multiple gene alterations via ML leads to better discrimination between sensitive and resistant PDXs in 19 of the 26 analysed cases. Highly predictive ML models employing concise gene lists were found for three cases: paclitaxel (breast cancer), binimetinib (breast cancer) and cetuximab (colorectal cancer). Interestingly, each of these multi-gene ML models identifies some treatment-responsive PDXs not harbouring the best actionable mutation for that case. Thus, ML multi-gene predictors generally have much fewer false negatives than the corresponding single-gene marker. (4) Conclusions: As PDXs often recapitulate clinical outcomes, these results suggest that many more patients could benefit from precision oncology if ML algorithms were also applied to existing clinical pharmacogenomics data, especially those algorithms generating classifiers combining data-selected gene alterations.

Drug sensitivity prediction from cell line-based pharmacogenomics data: guidelines for developing machine learning models.

The goal of precision oncology is to tailor treatment for patients individually using the genomic profile of their tumors. Pharmacogenomics datasets such as cancer cell lines are among the most valuable resources for drug sensitivity prediction, a crucial task of precision oncology. Machine learning methods have been employed to predict drug sensitivity based on the multiple omics data available for large panels of cancer cell lines. However, there are no comprehensive guidelines on how to properly train and validate such machine learning models for drug sensitivity prediction. In this paper, we introduce a set of guidelines for different aspects of training gene expression-based predictors using cell line datasets. These guidelines provide extensive analysis of the generalization of drug sensitivity predictors and challenge many current practices in the community including the choice of training dataset and measure of drug sensitivity. The application of these guidelines in future studies will enable the development of more robust preclinical biomarkers.

Abstract 1356: Meta-analysis and lack of independence assumption: Application in biomarker discovery

Abstract Due to the developments of high-throughput sequencing (HTS) technologies, massive amounts of data are currently available in cancer research which has provided a remarkable opportunity for the identification of biomarkers. However, biomarkers obtained from different studies of the same condition often show lack of agreement with each other due to concerns of inconsistency between laboratories in experimental designs and technical platforms for high throughput measurements and data processing computational methods. In addition, the large number of features and relatively small sample sizes can also lead findings from clinical and biological studies that are often not reproducible and can be biased when tested in independent cohorts. To address this issue, meta-analyses can be performed to reach more general and reliable conclusions along with increasing statistical power. In application, due to the limitations of experimental techniques and cost of HTS experiments, molecular data for some samples is not measured and is duplicated from available datasets in meta-analysis study. Therefore, in this scenario, the conventional meta-analysis procedures can be misleading when key assumptions such as independence of datasets or effect sizes are ignored. We assess the impact of ignoring the assumption of independence in applying the effect-size meta-analysis approaches to estimate heterogeneity across studies along with combined effect sizes using pharmacogenomic datasets. We considered molecular and drug responses data from Breast cancer and Pan-cancer cell lines sensitivity screenings obtained from the PharmacoGx package including CCLE, GDSC, GRAY, gCSI, CTRP and UHNBreast. Under each study, the meta preprocess including missing drug response imputation via multiple imputation by chained equations method using classification and regression trees, gene matching or filtering, and scaling or normalization are considered. For each study, the association between drug sensitivity and expression data is assessed by fitting (adjusted) linear regression models to identify cancer biomarkers. We evaluate the deviation from the assumption of independence by applying the integration analyses using non-independent effect sizes (i.e., standardized regression coefficient) related to the duplicated expression data. In addition, the relationship between increasing in the number of duplicated expression data and violation of non-independent effect-sizes is studied. The results indicate that duplicating gene expression data, therefore violating the independence assumption, can substantially increase the deviation of meta-estimates of effect sizes. Citation Format: Farnoosh Abbas-Aghababazadeh, Benjamin Haibe-Kains. Meta-analysis and lack of independence assumption: Application in biomarker discovery [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1356.

DeepDRK: a deep learning framework for drug repurposing through kernel-based multi-omics integration.

Recent pharmacogenomic studies that generate sequencing data coupled with pharmacological characteristics for patient-derived cancer cell lines led to large amounts of multi-omics data for precision cancer medicine. Among various obstacles hindering clinical translation, lacking effective methods for multimodal and multisource data integration is becoming a bottleneck. Here we proposed DeepDRK, a machine learning framework for deciphering drug response through kernel-based data integration. To transfer information among different drugs and cancer types, we trained deep neural networks on more than 20000 pan-cancer cell line-anticancer drug pairs. These pairs were characterized by kernel-based similarity matrices integrating multisource and multi-omics data including genomics, transcriptomics, epigenomics, chemical properties of compounds and known drug-target interactions. Applied to benchmark cancer cell line datasets, our model surpassed previous approaches with higher accuracy and better robustness. Then we applied our model on newly established patient-derived cancer cell lines and achieved satisfactory performance with AUC of 0.84 and AUPRC of 0.77. Moreover, DeepDRK was used to predict clinical response of cancer patients. Notably, the prediction of DeepDRK correlated well with clinical outcome of patients and revealed multiple drug repurposing candidates. In sum, DeepDRK provided a computational method to predict drug response of cancer cells from integrating pharmacogenomic datasets, offering an alternative way to prioritize repurposing drugs in precision cancer treatment. The DeepDRK is freely available via https://github.com/wangyc82/DeepDRK.

Bipartite graph-based approach for clustering of cell lines by gene expression-drug response associations.

In pharmacogenomic studies, the biological context of cell lines influences the predictive ability of drug-response models and the discovery of biomarkers. Thus, similar cell lines are often studied together based on prior knowledge of biological annotations. However, this selection approach is not scalable with the number of annotations, and the relationship between gene-drug association patterns and biological context may not be obvious. We present a procedure to compare cell lines based on their gene-drug association patterns. Starting with a grouping of cell lines from biological annotation, we model gene-drug association patterns for each group as a bipartite graph between genes and drugs. This is accomplished by applying sparse canonical correlation analysis (SCCA) to extract the gene-drug associations, and using the canonical vectors to construct the edge weights. Then, we introduce a nuclear norm-based dissimilarity measure to compare the bipartite graphs. Accompanying our procedure is a permutation test to evaluate the significance of similarity of cell line groups in terms of gene-drug associations. In the pharmacogenomic datasets CTRP2, GDSC2 and CCLE, hierarchical clustering of carcinoma groups based on this dissimilarity measure uniquely reveals clustering patterns driven by carcinoma subtype rather than primary site. Next, we show that the top associated drugs or genes from SCCA can be used to characterize the clustering patterns of haematopoietic and lymphoid malignancies. Finally, we confirm by simulation that when drug responses are linearly dependent on expression, our approach is the only one that can effectively infer the true hierarchy compared to existing approaches. Bipartite graph-based hierarchical clustering is implemented in R and can be obtained from CRAN: https://CRAN.R-project.org/package=hierBipartite. The source code is available at https://github.com/CalvinTChi/hierBipartite. The datasets were derived from sources in the public domain, which are the Cancer Cell Line Encyclopedia (https://portals.broadinstitute.org/ccle), the Cancer Therapeutics Response Portal (https://portals.broadinstitute.org/ctrp.v2.1/?page=#ctd2BodyHome), and the Genomics of Drug Sensitivity in Cancer (https://www.cancerrxgene.org/). These datasets can be downloaded using the PharmacoGx R package (https://bioconductor.org/packages/release/bioc/html/PharmacoGx.html). Supplementary data are available at Bioinformatics online.

Abstract PO-070: Bimodality of gene expression in cancer patient tumors as interpretable biomarkers for drug sensitivity

Abstract Identifying biomarkers predictive of cancer cells’ response to drug treatment constitutes one of the main challenges in precision oncology. Recent large-scale cancer pharmacogenomic studies have boosted the research for finding predictive biomarkers by profiling thousands of human cancer cell lines at the molecular level and screening them with hundreds of approved drugs and experimental chemical compounds. Many studies have leveraged these data to build predictive models of response using various statistical and machine learning methods. However, a common challenge in these methods is the lack of interpretability as to how they make the predictions and which features were the most associated with response, hindering the clinical translation of these models. To alleviate this issue, we develop a new machine learning pipeline based on the recent LOBICO approach that explores the space of bimodally expressed genes in multiple large in vitro pharmacogenomic studies and builds multivariate, nonlinear, yet interpretable logic-based models predictive of drug response. Using our method, we used a compendium of three of the largest pharmacogenomic data sets to build robust and interpretable models for 101 drugs that span 17 drug classes with high validation rate in independent datasets. Citation Format: Wail Ba-alawi, Sisira Kadambat Nair, Bo Li, Anthony Mammoliti, Petr Smirnov, Arvind Singh Mer, Linda Penn, Benjamin Haibe-Kains. Bimodality of gene expression in cancer patient tumors as interpretable biomarkers for drug sensitivity [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-070.