Biology subjects

Szczurek, E.

Publications and source records attributed to Szczurek, E..

25 records · Page 2Linked to original sources

bioRxiv2021

Interpretable deep recommender system model for prediction of kinase inhibitor efficacy across cancer cell lines

Computational models for drug sensitivity prediction have the potential to revolutionise personalized cancer medicine. Drug sensitivity assays, as well as profiling of cancer cell lines and drugs becomes increasingly available for training such models. Machine learning methods for drug sensitivity prediction must be optimized for: (i) leveraging the wealth of information about both cancer cell lines and drugs, (ii) predictive performance and (iii) interpretability. Multiple methods were proposed for predicting drug sensitivity from cancer cell line features, some in a multi-task fashion. So far, no such model leveraged drug inhibition profiles. Recent neural network-based recommender systems arise as models capable of predicting cancer cell line response to drugs from their biological features with high prediction accuracy. These models, however, require a tailored approach to model interpretability. In this work, we develop a neural network recommender system for kinase inhibitor sensitivity prediction called DEERS. The model utilizes molecular features of the cancer cell lines and kinase inhibition profiles of the drugs. DEERS incorporates two autoencoders to project cell line and drug features into 10-dimensional hidden representations and a feed-forward neural network to combine them into response prediction. We propose a novel model interpretability approach offering the widest possible assessment of the specific genes and biological processes that underlie the action of the drugs on the cell lines. The approach considers also such genes and processes that were not included in the set of modeled features. Our approach outperforms simpler matrix factorization models, achieving R=0.82 correlation between true and predicted response for the unseen cell lines. Using the interpretability analysis, we evaluate correlation of all human genes with each of the hidden cell line dimensions. Subsequently, we identify 67 biological processes associated with these dimensions. Combined with drug response data, these associations point at the processes that drive the cell line sensitivity to particular compounds. Detailed case studies are shown for PHA-793887, XMD14-99 and Dabrafenib. Our framework provides an expressive, multitask neural network model with a custom interpretability approach for inferring underlying biological factors and explaining cancer cell response to drugs.

bioinformatics↗

bioRxiv2020

Learning epistatic gene interactions from perturbation screens

AO_SCPLOWBSTRACTC_SCPLOWThe treatment of complex diseases often relies on combinatorial therapy, a strategy where drugs are used to target multiple genes simultaneously. Promising candidate genes for combinatorial perturbation often constitute epistatic genes, i.e., genes which contribute to a phenotype in a non-linear fashion. Experimental identification of the full landscape of genetic interactions by perturbing all gene combinations is prohibitive due to the exponential growth of testable hypotheses. Here we present a model for the inference of pairwise epistatic, including synthetic lethal, gene interactions from siRNA-based perturbation screens. The model exploits the combinatorial nature of siRNA-based screens resulting from the high numbers of sequence-dependent off-target effects, where each siRNA apart from its intended target knocks down hundreds of additional genes. We show that conditional and marginal epistasis can be estimated as interaction coefficients of regression models on perturbation data. We compare two methods, namely glinternet and xyz, for selecting non-zero effects in high dimensions as components of the model, and make recommendations for the appropriate use of each. For data simulated from real RNAi screening libraries, we show that glinternet successfully identifies epistatic gene pairs with high accuracy across a wide range of relevant parameters for the signal-to-noise ratio of observed phenotypes, the effect size of epistasis and the number of observations per double knockdown. xyz is also able to identify interactions from lower dimensional data sets (fewer genes), but is less accurate for many dimensions. Higher accuracy of glinternet, however, comes at the cost of longer running time compared to xyz. The general model is widely applicable and allows mining the wealth of publicly available RNAi screening data for the estimation of epistatic interactions between genes. As a proof of concept, we apply the model to search for interactions, and potential targets for treatment, among previously published sets of siRNA perturbation screens on various pathogens. The identified interactions include both known epistatic interactions as well as novel findings.

bioinformatics↗

bioRxiv2020

CACTUS: integrating clonal architecture with genomic clustering and transcriptome profiling of single tumor cells

BackgroundDrawing genotype-to-phenotype maps in tumors is of paramount importance for understanding tumor heterogeneity. Assignment of single cells to their tumor clones of origin can be approached by matching the genotypes of the clones to the mutations found in RNA sequencing of the cells. The confidence of the cell-to-clone mapping can be increased by accounting for additional measurements. Follicular lymphoma, a malignancy of mature B cells that continuously acquire mutations in parallel in the exome and in B-cell receptor loci, presents a unique opportunity to align exome-derived mutations with B-cell receptor clonotypes as an independent measure for clonal evolution. ResultsHere, we propose CACTUS, a probabilistic model that leverages the information from an independent genomic clustering of cells and exploits the scarce single cell RNA sequencing data to map single cells to given imperfect genotypes of tumor clones. We apply CACTUS to two follicular lymphoma patient samples, integrating three measurements: whole exome sequencing, single cell RNA sequencing, and B-cell receptor sequencing. CACTUS outperforms a predecessor model by confidently assigning cells and B-cell receptor clonotypes to the tumor clones. ConclusionsThe integration of independent measurements increases model certainty and is the key to improving model performance in the challenging task of charting the genotype-to-phenotype maps in tumors. CACTUS opens the avenue to study the functional implications of tumor heterogeneity, and origins of resistance to targeted therapies.

cancer biology↗

bioRxiv2019

Feature selection strategies for drug sensitivity prediction

Drug sensitivity prediction constitutes one of the main challenges in personalized medicine. The major difficulty of this problem stems from the fact that the sensitivity of cancer cells to treatment depends on an unknown subset of a large number of biological features. Although feature selection is the key to interpretable results and identification of potential biomarkers, a comprehensive assessment of feature selection methods for drug sensitivity prediction has so far not been performed. We propose feature selection approaches driven by prior knowledge of drug targets, target pathways, and gene expression signatures. We asses these methodologies on Genomics of Drug Sensitivity in Cancer (GDSC) dataset, a panel of around 1000 cell lines screened against multiple anticancer compounds. We compare our results with a baseline model utilizing genome-wide gene expression features and common data-driven feature selection techniques. Together, 2484 unique models were evaluated, providing a comprehensive study of feature selection strategies for the drug response prediction problem. For 23 drugs, the models achieve better predictive performance when the features are selected according to prior knowledge of drug targets and pathways. The best correlation of observed and predicted response using the test set is achieved for Linifanib (r=0.75). Extending the drug-dependent features with gene expression signatures yields models that are most predictive of drug response for 60 drugs, with the best performing example of Dabrafenib. Examples of how pre-selection of features benefits the model interpretability are given for Dabrafenib, Linifanib and Quizartinib. Based on GDSC drug data, we find that feature selection driven by prior knowledge tends to yield better results for drugs targeting specific genes and pathways, while models with the genome-wide features perform better for drugs affecting general mechanisms such as metabolism and DNA replication. For a significant group of the compounds, even a very small number of features based on simple drug properties is often highly predictive of drug sensitivity, can explain the mechanism of drug action and be used as guidelines for their prescription. In general, choosing appropriate feature selection strategies has the potential to develop interpretable models that are indicative for therapy design.

bioinformatics↗

bioRxiv2019

ARA: accurate, reliable and active histopathological image classification framework with Bayesian deep learning

Machine learning algorithms hold the promise to effectively automate the analysis of histopathological images that are routinely generated in clinical practice. Any machine learning method used in the clinical diagnostic process has to be extremely accurate and, ideally, provide a measure of uncertainty for its predictions. Such accurate and reliable classifiers need enough labelled data for training, which requires time-consuming and costly manual annotation by pathologists. Thus, it is critical to minimise the amount of data needed to reach the desired accuracy by maximising the efficiency of training. We propose an accurate, reliable and active (ARA) image classification framework and introduce a new Bayesian Convolutional Neural Network (ARA-CNN) for classifying histopathological images of colorectal cancer. The model achieves exceptional classification accuracy, outperforming other models trained on the same dataset. The network outputs an uncertainty measurement for each tested image. We show that uncertainty measures can be used to detect mislabelled training samples and can be employed in an efficient active learning workflow. Using a variational dropout-based entropy measure of uncertainty in the workflow speeds up the learning process by roughly 45%. Finally, we utilise our model to segment whole-slide images of colorectal tissue and compute segmentation-based spatial statistics.

bioinformatics↗

bioRxiv2018

A stochastic model of metastatic bottleneck predicts patient outcome and therapy response

Metastases are responsible for 90% of cancer-related deaths. Initiation of metastases, where newly seeded tumor cells expand into colonies, presents a tremendous bottleneck to metastasis formation. Despite its clinical importance, our understanding of this process is very limited. Here, we propose a simple stochastic model assuming that the initiating metastatic cells proliferate faster when surrounded by more of their kind. The model quantifies the severity of metastatic bottleneck as the probability that the seeded colony survives. Based on this model, we derive how metastasis occurrence depends on primary tumor size and affects patient outcome. Our predictions agree with epidemiological data for thirteen cancer types. The model predicts that impact of treatment decisions depends both on the primary tumor size and on the severity of the metastatic bottleneck, and that medical interventions that tighten the bottleneck would be much more efficient than therapies that decrease overall tumor burden, such as chemotherapy.

cancer biology↗

bioRxiv2017

Epistasis in genomic and survival data of cancer patients

Cancer aggressiveness and its effect on patient survival depends on mutations in the tumor genome. Epistatic interactions between the mutated genes may guide the choice of anticancer therapy and set predictive factors of its success. Inhibitors targeting synthetic lethal partners of genes mutated in tumors are already utilized for efficient and specific treatment in the clinic. The space of possible epistatic interactions, how-ever, is overwhelming, and computational methods are needed to limit the experimental effort of validating the interactions for therapy and characterizing their biomarkers. Here, we introduce SurvLRT, a statistical likelihood ratio test for identifying epistatic gene pairs and triplets from cancer patient genomic and survival data. Compared to established approaches, SurvLRT performed favorable in predicting known, experimentally verified synthetic lethal partners of PARP1 from TCGA data. Our approach is the first to test for epistasis between triplets of genes to identify biomarkers of synthetic lethality-based therapy. SurvLRT proved successful in identifying the known gene TP53BP1 as the biomarker of success of the therapy targeting PARP in BRCA1 deficient tumors. Search for other biomarkers for the same interaction revealed a region whose deletion was a more significant biomarker than deletion of TP53BP1. With the ability to detect not only pairwise but twelve different types of triple epistasis, applicability of SurvLRT goes beyond cancer therapy, to the level of characterization of shapes of fitness landscapes.\n\nAuthor SummaryGenomic alterations in tumors affect the fitness of tumor cells, controlling how well they replicate and survive compared to other cells. The landscape of tumor fitness is shaped by epistasis. Epistasis occurs when the contribution of gene alterations to the total fitness is non-linear. The type of epistatic genetic interactions with great potential for cancer therapy is synthetic lethality. Inhibitors targeting synthetic lethal partners of genes mutated in tumors can selectively kill tumor and not normal cells. Therapy based on synthetic lethality is, however, context dependent, and it is crucial to identify its biomarkers. Unfortunately, the space of possible interactions and their biomarkers is overwhelming for experimental validation. Computational pre-selection methods are required to limit the experimental effort. Here, we introduce a statistical approach called SurvLRT, for the identification of epistatic gene pairs and triplets based on patient genomic and survival data. First, we show that using SurvLRT, we can deliver synthetic lethal interactions of pairs of genes that are specific to cancer. Second, we demonstrate the applicability of SurvLRT to identify biomarkers for synthetic lethality, such as mutational status of other genes that can alleviate the synthetic effect.

genomics↗