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

Reliable anti-cancer drug sensitivity prediction and prioritization

The application of machine learning (ML) to solve real-world problems does not only bear great potential but also high risk. One fundamental challenge in risk mitigation is to ensure the reliability of the ML predictions, i.e., the model error should be minimized, and the prediction uncertainty should be estimated. Especially for medical applications, the importance of reliable predictions can not be understated. Here, we address this challenge for anti-cancer drug sensitivity prediction and prioritization. To this end, we present a novel drug sensitivity prediction and prioritization approach guaranteeing user-specified certainty levels. The developed conformal prediction approach is applicable to classification, regression, and simultaneous regression and classification. Additionally, we propose a novel drug sensitivity measure that is based on clinically relevant drug concentrations and enables a straightforward prioritization of drugs for a given cancer sample.

Overcoming limitations in current measures of drug response may enable AI-driven precision oncology

Machine learning (ML) models of drug sensitivity prediction are becoming increasingly popular in precision oncology. Here, we identify a fundamental limitation in standard measures of drug sensitivity that hinders the development of personalized prediction models - they focus on absolute effects but do not capture relative differences between cancer subtypes. Our work suggests that using z-scored drug response measures mitigates these limitations and leads to meaningful predictions, opening the door for sophisticated ML precision oncology models.

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

Motivation

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.

Results

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.

Availability and implementation

https://github.com/merlab/text_features.

MMDRP: drug response prediction and biomarker discovery using multi-modal deep learning

Motivation

A major challenge in cancer care is that patients with similar demographics, tumor types, and medical histories can respond quite differently to the same drug regimens. This difference is largely explained by genetic and other molecular variabilities among the patients and their cancers. Efforts in the pharmacogenomics field are underway to understand better the relationship between the genome of the patient's healthy and tumor cells and their response to therapy. To advance this goal, research groups and consortia have undertaken large-scale systematic screening of panels of drugs across multiple cancer cell lines that have been molecularly profiled by genomics, proteomics, and similar techniques. These large data drug screening sets have been applied to the problem of drug response prediction (DRP), the challenge of predicting the response of a previously untested drug/cell-line combination. Although deep learning algorithms outperform traditional methods, there are still many challenges in DRP that ultimately result in these models' low generalizability and hampers their clinical application.

Results

In this article, we describe a novel algorithm that addresses the major shortcomings of current DRP methods by combining multiple cell line characterization data, addressing drug response data skewness, and improving chemical compound representation.

Availability and implementation

MMDRP is implemented as an open-source, Python-based, command-line program and is available at https://github.com/LincolnSteinLab/MMDRP.

Comparison of multiple modalities for drug response prediction with learning curves using neural networks and XGBoost

Motivation

Anti-cancer drug response prediction is a central problem within stratified medicine. Transcriptomic profiles of cancer cell lines are typically used for drug response prediction, but we hypothesize that proteomics or phosphoproteomics might be more suitable as they give a more direct insight into cellular processes. However, there has not yet been a systematic comparison between all three of these datatypes using consistent evaluation criteria.

Results

Due to the limited number of cell lines with phosphoproteomics profiles we use learning curves, a plot of predictive performance as a function of dataset size, to compare the current performance and predict the future performance of the three omics datasets with more data. We use neural networks and XGBoost and compare them against a simple rule-based benchmark. We show that phosphoproteomics slightly outperforms RNA-seq and proteomics using the 38 cell lines with profiles of all three omics data types. Furthermore, using the 877 cell lines with proteomics and RNA-seq profiles, we show that RNA-seq slightly outperforms proteomics. With the learning curves we predict that the mean squared error using the phosphoproteomics dataset would decrease by ∼ 15 % if a dataset of the same size as the proteomics/transcriptomics was collected. For the cell lines with proteomics and RNA-seq profiles the learning curves reveal that for smaller dataset sizes neural networks outperform XGBoost and vice versa for larger datasets. Furthermore, the trajectory of the XGBoost curve suggests that it will improve faster than the neural networks as more data are collected.

Availability and implementation

See https://github.com/Nik-BB/Learning-curves-for-DRP for the code used.

Towards artificial intelligence-enabled extracellular vesicle precision drug delivery

Extracellular Vesicles (EVs), particularly exosomes, recently exploded into nanomedicine as an emerging drug delivery approach due to their superior biocompatibility, circulating stability, and bioavailability in vivo. However, EV heterogeneity makes molecular targeting precision a critical challenge. Deciphering key molecular drivers for controlling EV tissue targeting specificity is in great need. Artificial intelligence (AI) brings powerful prediction ability for guiding the rational design of engineered EVs in precision control for drug delivery. This review focuses on cutting-edge nano-delivery via integrating large-scale EV data with AI to develop AI-directed EV therapies and illuminate the clinical translation potential. We briefly review the current status of EVs in drug delivery, including the current frontier, limitations, and considerations to advance the field. Subsequently, we detail the future of AI in drug delivery and its impact on precision EV delivery. Our review discusses the current universal challenge of standardization and critical considerations when using AI combined with EVs for precision drug delivery. Finally, we will conclude this review with a perspective on future clinical translation led by a combined effort of AI and EV research.

NeuPD-A Neural Network-Based Approach to Predict Antineoplastic Drug Response

With the beginning of the high-throughput screening, in silico-based drug response analysis has opened lots of research avenues in the field of personalized medicine. For a decade, many different predicting techniques have been recommended for the antineoplastic (anti-cancer) drug response, but still, there is a need for improvements in drug sensitivity prediction. The intent of this research study is to propose a framework, namely NeuPD, to validate the potential anti-cancer drugs against a panel of cancer cell lines in publicly available datasets. The datasets used in this work are Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Cell Line Encyclopedia (CCLE). As not all drugs are effective on cancer cell lines, we have worked on 10 essential drugs from the GDSC dataset that have achieved the best modeling results in previous studies. We also extracted 1610 essential oncogene expressions from 983 cell lines from the same dataset. Whereas, from the CCLE dataset, 16,383 gene expressions from 1037 cell lines and 24 drugs have been used in our experiments. For dimensionality reduction, Pearson correlation is applied to best fit the model. We integrate the genomic features of cell lines and drugs' fingerprints to fit the neural network model. For evaluation of the proposed NeuPD framework, we have used repeated K-fold cross-validation with 5 times repeats where K = 10 to demonstrate the performance in terms of root mean square error (RMSE) and coefficient determination (R²). The results obtained on the GDSC dataset that were measured using these cost functions show that our proposed NeuPD framework has outperformed existing approaches with an RMSE of 0.490 and R² of 0.929.

Leveraging multi-source to resolve inconsistency across pharmacogenomic datasets in drug sensitivity prediction

Pharmacogenomics datasets have been generated for various purposes, such as investigating different biomarkers. However, when studying the same cell line with the same drugs, differences in drug responses exist between studies. These variations arise from factors such as inter-tumoral heterogeneity, experimental standardization, and the complexity of cell subtypes. Consequently, drug response prediction suffers from limited generalizability. To address these challenges, we propose a computational model based on Federated Learning (FL) for drug response prediction. By leveraging three pharmacogenomics datasets (CCLE, GDSC2, and gCSI), we evaluate the performance of our model across diverse cell line-based databases. Our results demonstrate superior predictive performance compared to baseline methods and traditional FL approaches through various experimental tests. This study underscores the potential of employing FL to leverage multiple data sources, enabling the development of generalized models that account for inconsistencies among pharmacogenomics datasets. By addressing the limitations of low generalizability, our approach contributes to advancing drug response prediction in precision oncology.

Small molecule-mediated targeting of microRNAs for drug discovery: Experiments, computational techniques, and disease implications

Small molecules have been providing medical breakthroughs for human diseases for more than a century. Recently, identifying small molecule inhibitors that target microRNAs (miRNAs) has gained importance, despite the challenges posed by labour-intensive screening experiments and the significant efforts required for medicinal chemistry optimization. Numerous experimentally-verified cases have demonstrated the potential of miRNA-targeted small molecule inhibitors for disease treatment. This new approach is grounded in their posttranscriptional regulation of the expression of disease-associated genes. Reversing dysregulated gene expression using this mechanism may help control dysfunctional pathways. Furthermore, the ongoing improvement of algorithms has allowed for the integration of computational strategies built on top of laboratory-based data, facilitating a more precise and rational design and discovery of lead compounds. To complement the use of extensive pharmacogenomics data in prioritising potential drugs, our previous work introduced a computational approach based on only molecular sequences. Moreover, various computational tools for predicting molecular interactions in biological networks using similarity-based inference techniques have been accumulated in established studies. However, there are a limited number of comprehensive reviews covering both computational and experimental drug discovery processes. In this review, we outline a cohesive overview of both biological and computational applications in miRNA-targeted drug discovery, along with their disease implications and clinical significance. Finally, utilizing drug-target interaction (DTIs) data from DrugBank, we showcase the effectiveness of deep learning for obtaining the physicochemical characterization of DTIs.

Deep learning methods for drug response prediction in cancer: Predominant and emerging trends

Cancer claims millions of lives yearly worldwide. While many therapies have been made available in recent years, by in large cancer remains unsolved. Exploiting computational predictive models to study and treat cancer holds great promise in improving drug development and personalized design of treatment plans, ultimately suppressing tumors, alleviating suffering, and prolonging lives of patients. A wave of recent papers demonstrates promising results in predicting cancer response to drug treatments while utilizing deep learning methods. These papers investigate diverse data representations, neural network architectures, learning methodologies, and evaluations schemes. However, deciphering promising predominant and emerging trends is difficult due to the variety of explored methods and lack of standardized framework for comparing drug response prediction models. To obtain a comprehensive landscape of deep learning methods, we conducted an extensive search and analysis of deep learning models that predict the response to single drug treatments. A total of 61 deep learning-based models have been curated, and summary plots were generated. Based on the analysis, observable patterns and prevalence of methods have been revealed. This review allows to better understand the current state of the field and identify major challenges and promising solution paths.

A systematic assessment of deep learning methods for drug response prediction: from in vitro to clinical applications

Drug response prediction is an important problem in personalized cancer therapy. Among various newly developed models, significant improvement in prediction performance has been reported using deep learning methods. However, systematic comparisons of deep learning methods, especially of the transferability from preclinical models to clinical cohorts, are currently lacking. To provide a more rigorous assessment, the performance of six representative deep learning methods for drug response prediction using nine evaluation metrics, including the overall prediction accuracy, predictability of each drug, potential associated factors and transferability to clinical cohorts, in multiple application scenarios was benchmarked. Most methods show promising prediction within cell line datasets, and TGSA, with its lower time cost and better performance, is recommended. Although the performance metrics decrease when applying models trained on cell lines to patients, a certain amount of power to distinguish clinical response on some drugs can be maintained using CRDNN and TGSA. With these assessments, we provide a guidance for researchers to choose appropriate methods, as well as insights into future directions for the development of more effective methods in clinical scenarios.

Data augmentation and multimodal learning for predicting drug response in patient-derived xenografts from gene expressions and histology images

Patient-derived xenografts (PDXs) are an appealing platform for preclinical drug studies. A primary challenge in modeling drug response prediction (DRP) with PDXs and neural networks (NNs) is the limited number of drug response samples. We investigate multimodal neural network (MM-Net) and data augmentation for DRP in PDXs. The MM-Net learns to predict response using drug descriptors, gene expressions (GE), and histology whole-slide images (WSIs). We explore whether combining WSIs with GE improves predictions as compared with models that use GE alone. We propose two data augmentation methods which allow us training multimodal and unimodal NNs without changing architectures with a single larger dataset: 1) combine single-drug and drug-pair treatments by homogenizing drug representations, and 2) augment drug-pairs which doubles the sample size of all drug-pair samples. Unimodal NNs which use GE are compared to assess the contribution of data augmentation. The NN that uses the original and the augmented drug-pair treatments as well as single-drug treatments outperforms NNs that ignore either the augmented drug-pairs or the single-drug treatments. In assessing the multimodal learning based on the MCC metric, MM-Net outperforms all the baselines. Our results show that data augmentation and integration of histology images with GE can improve prediction performance of drug response in PDXs.

Utilization of Cancer Cell Line Screening to Elucidate the Anticancer Activity and Biological Pathways Related to the Ruthenium-Based Therapeutic BOLD-100

BOLD-100 (sodium trans-[tetrachlorobis(1H indazole)ruthenate(III)]) is a ruthenium-based anticancer compound currently in clinical development. The identification of cancer types that show increased sensitivity towards BOLD-100 can lead to improved developmental strategies. Sensitivity profiling can also identify mechanisms of action that are pertinent for the bioactivity of complex therapeutics. Sensitivity to BOLD-100 was measured in a 319-cancer-cell line panel spanning 24 tissues. BOLD-100's sensitivity profile showed variation across the tissue lineages, including increased response in esophageal, bladder, and hematologic cancers. Multiple cancers, including esophageal, bile duct and colon cancer, had higher relative response to BOLD-100 than to cisplatin. Response to BOLD-100 showed only moderate correlation to anticancer compounds in the Genomics of Drug Sensitivity in Cancer (GDSC) database, as well as no clear theme in bioactivity of correlated hits, suggesting that BOLD-100 may have a differentiated therapeutic profile. The genomic modalities of cancer cell lines were modeled against the BOLD-100 sensitivity profile, which revealed that genes related to ribosomal processes were associated with sensitivity to BOLD-100. Machine learning modeling of the sensitivity profile to BOLD-100 and gene expression data provided moderative predictive value. These findings provide further mechanistic understanding around BOLD-100 and support its development for additional cancer 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.

CREAMMIST: an integrative probabilistic database for cancer drug response prediction

Extensive in vitro cancer drug screening datasets have enabled scientists to identify biomarkers and develop machine learning models for predicting drug sensitivity. While most advancements have focused on omics profiles, cancer drug sensitivity scores precalculated by the original sources are often used as-is, without consideration for variabilities between studies. It is well-known that significant inconsistencies exist between the drug sensitivity scores across datasets due to differences in experimental setups and preprocessing methods used to obtain the sensitivity scores. As a result, many studies opt to focus only on a single dataset, leading to underutilization of available data and a limited interpretation of cancer pharmacogenomics analysis. To overcome these caveats, we have developed CREAMMIST (https://creammist.mtms.dev), an integrative database that enables users to obtain an integrative dose-response curve, to capture uncertainty (or high certainty when multiple datasets well align) across five widely used cancer cell-line drug-response datasets. We utilized the Bayesian framework to systematically integrate all available dose-response values across datasets (>14 millions dose-response data points). CREAMMIST provides easy-to-use statistics derived from the integrative dose-response curves for various downstream analyses such as identifying biomarkers, selecting drug concentrations for experiments, and training robust machine learning models.

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

Current 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.

•

Pan-cancer cell lines provide a test bench for exploring gene-drug relationships

•

Multi-omics data were integrated with pharmacological profiles for joint modeling

•

Mix-lasso identifies tissue-specific biomarkers predictive of multi-drug responses

•

Mix-lasso provides small number of stable features for drug discovery applications

Construction and Validation of a UPR-Associated Gene Prognostic Model for Head and Neck Squamous Cell Carcinoma

Our study is aimed at constructing and validating a UPR-associated gene signature to predict HNSCC prognosis. We obtained 544 samples of RNA sequencing data and clinical characteristics from TCGA database and randomly grouped the samples into training and testing cohorts (1 : 1 ratio). After identifying 14 UPR-associated genes with LASSO and univariate Cox regression analysis, HNSCC samples were categorized into low-risk (LR) and high-risk (HR) subgroups depending on the risk score. Our analyses indicated that low-risk patients had a much better prognosis in the training and testing cohorts. To predict the HNSCC prognosis with the 14 UPR-associated gene signatures, we incorporated the UPR gene risk score, N stage, M stage, and age into a nomogram model. We further explored the sensitivity to anticancer drugs by using the IC50 analysis in two subgroups from the Cancer Genome Project database. The outcomes showed that the AKT inhibitor III and sorafenib were sensitive anticancer drugs in HR and LR patients, respectively. The immune cell infiltration analysis and GSEA provided strong evidence for elucidating the molecular mechanisms of UPR-associated genes affecting HNSCC. In conclusion, the UPR-associated gene risk score, N stage, M stage, and age can serve as a robust model for predicting prognosis and can improve decision-making at the individual patient level.

Bimodal gene expression in cancer patients provides interpretable biomarkers for drug sensitivity

Identifying biomarkers predictive of cancer cell 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 pitfall to these methods is the lack of interpretability as to how they make 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. The performance of this approach was showcased 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 rates in independent datasets. These results along with in vivo and clinical validation, support a better translation of gene expression biomarkers between model systems using bimodal gene expression.