Phenotype prediction from single-cell RNA-seq data using attention-based neural networks.

MOTIVATION: A patient's disease phenotype can be driven and determined by specific groups of cells whose marker genes are either unknown or can only be detected at late-stage using conventional bulk assays such as RNA-Seq technology. Recent advances in single-cell RNA sequencing (scRNA-seq) enable gene expression profiling in cell-level resolution, and therefore have the potential to identify those cells driving the disease phenotype even while the number of these cells is small. However, most existing methods rely heavily on accurate cell type detection, and the number of available annotated samples is usually too small for training deep learning predictive models. RESULTS: Here, we propose the method ScRAT for phenotype prediction using scRNA-seq data. To train ScRAT with a limited number of samples of different phenotypes, such as coronavirus disease (COVID) and non-COVID, ScRAT first applies a mixup module to increase the number of training samples. A multi-head attention mechanism is employed to learn the most informative cells for each phenotype without relying on a given cell type annotation. Using three public COVID datasets, we show that ScRAT outperforms other phenotype prediction methods. The performance edge of ScRAT over its competitors increases as the number of training samples decreases, indicating the efficacy of our sample mixup. Critical cell types detected based on high-attention cells also support novel findings in the original papers and the recent literature. This suggests that ScRAT overcomes the challenge of missing marker genes and limited sample number with great potential revealing novel molecular mechanisms and/or therapies. AVAILABILITY AND IMPLEMENTATION: The code of our proposed method ScRAT is published at https://github.com/yuzhenmao/ScRAT.

Identifying Synergistic Components of Botanical Fungicide Formulations Using Interpretable Graph Neural Networks.

Botanical formulations are promising candidates for developing new biopesticides that can protect crops from pests and diseases while reducing harm to the environment. These biopesticides can be combined with permeation enhancer compounds to boost their efficacy against pests and fungal diseases. However, finding synergistic combinations of these compounds is challenging due to the large and complex chemical space. In this paper, we propose a novel deep learning method that can predict the synergy of botanical products and permeation enhancers based on in vitro assay data. Our method uses a weighted combination of component feature vectors to represent the input mixtures, which enables the model to handle a variable number of components and to interpret the contribution of each component to the synergy. We also employ an ensemble of interpretation methods to provide insights into the underlying mechanisms of synergy. We validate our method by testing the predicted synergistic combinations in wet-lab experiments and show that our method can discover novel and effective biopesticides that would otherwise be difficult to find. Our method is generalizable and applicable to other domains, where predicting mixtures of chemical compounds is important.

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.

Aristotle: stratified causal discovery for omics data.

BACKGROUND: There has been a simultaneous increase in demand and accessibility across genomics, transcriptomics, proteomics and metabolomics data, known as omics data. This has encouraged widespread application of omics data in life sciences, from personalized medicine to the discovery of underlying pathophysiology of diseases. Causal analysis of omics data may provide important insight into the underlying biological mechanisms. Existing causal analysis methods yield promising results when identifying potential general causes of an observed outcome based on omics data. However, they may fail to discover the causes specific to a particular stratum of individuals and missing from others. METHODS: To fill this gap, we introduce the problem of stratified causal discovery and propose a method, Aristotle, for solving it. Aristotle addresses the two challenges intrinsic to omics data: high dimensionality and hidden stratification. It employs existing biological knowledge and a state-of-the-art patient stratification method to tackle the above challenges and applies a quasi-experimental design method to each stratum to find stratum-specific potential causes. RESULTS: Evaluation based on synthetic data shows better performance for Aristotle in discovering true causes under different conditions compared to existing causal discovery methods. Experiments on a real dataset on Anthracycline Cardiotoxicity indicate that Aristotle's predictions are consistent with the existing literature. Moreover, Aristotle makes additional predictions that suggest further investigations.

Identification of differentially expressed gene modules in heterogeneous diseases.

MOTIVATION: Identification of differentially expressed genes is necessary for unraveling disease pathogenesis. This task is complicated by the fact that many diseases are heterogeneous at the molecular level and samples representing distinct disease subtypes may demonstrate different patterns of dysregulation. Biclustering methods are capable of identifying genes that follow a similar expression pattern only in a subset of samples and hence can consider disease heterogeneity. However, identifying biologically significant and reproducible sets of genes and samples remain challenging for the existing tools. Many recent studies have shown that the integration of gene expression and protein interaction data improves the robustness of prediction and classification and advances biomarker discovery. RESULTS: Here, we present DESMOND, a new method for identification of Differentially ExpreSsed gene MOdules iN Diseases. DESMOND performs network-constrained biclustering on gene expression data and identifies gene modules-connected sets of genes up- or down-regulated in subsets of samples. We applied DESMOND on expression profiles of samples from two large breast cancer cohorts and have shown that the capability of DESMOND to incorporate protein interactions allows identifying the biologically meaningful gene and sample subsets and improves the reproducibility of the results. AVAILABILITY AND IMPLEMENTATION: https://github.com/ozolotareva/DESMOND. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Ligand Binding Prediction Using Protein Structure Graphs and Residual Graph Attention Networks.

Computational prediction of ligand-target interactions is a crucial part of modern drug discovery as it helps to bypass high costs and labor demands of in vitro and in vivo screening. As the wealth of bioactivity data accumulates, it provides opportunities for the development of deep learning (DL) models with increasing predictive powers. Conventionally, such models were either limited to the use of very simplified representations of proteins or ineffective voxelization of their 3D structures. Herein, we present the development of the PSG-BAR (Protein Structure Graph-Binding Affinity Regression) approach that utilizes 3D structural information of the proteins along with 2D graph representations of ligands. The method also introduces attention scores to selectively weight protein regions that are most important for ligand binding. Results: The developed approach demonstrates the state-of-the-art performance on several binding affinity benchmarking datasets. The attention-based pooling of protein graphs enables identification of surface residues as critical residues for protein-ligand binding. Finally, we validate our model predictions against an experimental assay on a viral main protease (Mpro)-the hallmark target of SARS-CoV-2 coronavirus.

AITL: Adversarial Inductive Transfer Learning with input and output space adaptation for pharmacogenomics.

MOTIVATION: The goal of pharmacogenomics is to predict drug response in patients using their single- or multi-omics data. A major challenge is that clinical data (i.e. patients) with drug response outcome is very limited, creating a need for transfer learning to bridge the gap between large pre-clinical pharmacogenomics datasets (e.g. cancer cell lines), as a source domain, and clinical datasets as a target domain. Two major discrepancies exist between pre-clinical and clinical datasets: (i) in the input space, the gene expression data due to difference in the basic biology, and (ii) in the output space, the different measures of the drug response. Therefore, training a computational model on cell lines and testing it on patients violates the i.i.d assumption that train and test data are from the same distribution. RESULTS: We propose Adversarial Inductive Transfer Learning (AITL), a deep neural network method for addressing discrepancies in input and output space between the pre-clinical and clinical datasets. AITL takes gene expression of patients and cell lines as the input, employs adversarial domain adaptation and multi-task learning to address these discrepancies, and predicts the drug response as the output. To the best of our knowledge, AITL is the first adversarial inductive transfer learning method to address both input and output discrepancies. Experimental results indicate that AITL outperforms state-of-the-art pharmacogenomics and transfer learning baselines and may guide precision oncology more accurately. AVAILABILITY AND IMPLEMENTATION: https://github.com/hosseinshn/AITL. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOLI: multi-omics late integration with deep neural networks for drug response prediction.

MOTIVATION: Historically, gene expression has been shown to be the most informative data for drug response prediction. Recent evidence suggests that integrating additional omics can improve the prediction accuracy which raises the question of how to integrate the additional omics. Regardless of the integration strategy, clinical utility and translatability are crucial. Thus, we reasoned a multi-omics approach combined with clinical datasets would improve drug response prediction and clinical relevance. RESULTS: We propose MOLI, a multi-omics late integration method based on deep neural networks. MOLI takes somatic mutation, copy number aberration and gene expression data as input, and integrates them for drug response prediction. MOLI uses type-specific encoding sub-networks to learn features for each omics type, concatenates them into one representation and optimizes this representation via a combined cost function consisting of a triplet loss and a binary cross-entropy loss. The former makes the representations of responder samples more similar to each other and different from the non-responders, and the latter makes this representation predictive of the response values. We validate MOLI on in vitro and in vivo datasets for five chemotherapy agents and two targeted therapeutics. Compared to state-of-the-art single-omics and early integration multi-omics methods, MOLI achieves higher prediction accuracy in external validations. Moreover, a significant improvement in MOLI's performance is observed for targeted drugs when training on a pan-drug input, i.e. using all the drugs with the same target compared to training only on drug-specific inputs. MOLI's high predictive power suggests it may have utility in precision oncology. AVAILABILITY AND IMPLEMENTATION: https://github.com/hosseinshn/MOLI. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Collaborative intra-tumor heterogeneity detection.

MOTIVATION: Despite the remarkable advances in sequencing and computational techniques, noise in the data and complexity of the underlying biological mechanisms render deconvolution of the phylogenetic relationships between cancer mutations difficult. Besides that, the majority of the existing datasets consist of bulk sequencing data of single tumor sample of an individual. Accurate inference of the phylogenetic order of mutations is particularly challenging in these cases and the existing methods are faced with several theoretical limitations. To overcome these limitations, new methods are required for integrating and harnessing the full potential of the existing data. RESULTS: We introduce a method called Hintra for intra-tumor heterogeneity detection. Hintra integrates sequencing data for a cohort of tumors and infers tumor phylogeny for each individual based on the evolutionary information shared between different tumors. Through an iterative process, Hintra learns the repeating evolutionary patterns and uses this information for resolving the phylogenetic ambiguities of individual tumors. The results of synthetic experiments show an improved performance compared to two state-of-the-art methods. The experimental results with a recent Breast Cancer dataset are consistent with the existing knowledge and provide potentially interesting findings. AVAILABILITY AND IMPLEMENTATION: The source code for Hintra is available at https://github.com/sahandk/HINTRA.

SUBSTRA: Supervised Bayesian Patient Stratification.

MOTIVATION: Patient stratification methods are key to the vision of precision medicine. Here, we consider transcriptional data to segment the patient population into subsets relevant to a given phenotype. Whereas most existing patient stratification methods focus either on predictive performance or interpretable features, we developed a method striking a balance between these two important goals. RESULTS: We introduce a Bayesian method called SUBSTRA that uses regularized biclustering to identify patient subtypes and interpretable subtype-specific transcript clusters. The method iteratively re-weights feature importance to optimize phenotype prediction performance by producing more phenotype-relevant patient subtypes. We investigate the performance of SUBSTRA in finding relevant features using simulated data and successfully benchmark it against state-of-the-art unsupervised stratification methods and supervised alternatives. Moreover, SUBSTRA achieves predictive performance competitive with the supervised benchmark methods and provides interpretable transcriptional features in diverse biological settings, such as drug response prediction, cancer diagnosis, or kidney transplant rejection. AVAILABILITY AND IMPLEMENTATION: The R code of SUBSTRA is available at https://github.com/sahandk/SUBSTRA. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Uncovering the subtype-specific temporal order of cancer pathway dysregulation.

Cancer is driven by genetic mutations that dysregulate pathways important for proper cell function. Therefore, discovering these cancer pathways and their dysregulation order is key to understanding and treating cancer. However, the heterogeneity of mutations between different individuals makes this challenging and requires that cancer progression is studied in a subtype-specific way. To address this challenge, we provide a mathematical model, called Subtype-specific Pathway Linear Progression Model (SPM), that simultaneously captures cancer subtypes and pathways and order of dysregulation of the pathways within each subtype. Experiments with synthetic data indicate the robustness of SPM to problem specifics including noise compared to an existing method. Moreover, experimental results on glioblastoma multiforme and colorectal adenocarcinoma show the consistency of SPM's results with the existing knowledge and its superiority to an existing method in certain cases. The implementation of our method is available at https://github.com/Dalton386/SPM.

Deep Learning Modeling of Androgen Receptor Responses to Prostate Cancer Therapies.

Gain-of-function mutations in human androgen receptor (AR) are among the major causes of drug resistance in prostate cancer (PCa). Identifying mutations that cause resistant phenotype is of critical importance for guiding treatment protocols, as well as for designing drugs that do not elicit adverse responses. However, experimental characterization of these mutations is time consuming and costly; thus, predictive models are needed to anticipate resistant mutations and to guide the drug discovery process. In this work, we leverage experimental data collected on 68 AR mutants, either observed in the clinic or described in the literature, to train a deep neural network (DNN) that predicts the response of these mutants to currently used and experimental anti-androgens and testosterone. We demonstrate that the use of this DNN, with general 2D descriptors, provides a more accurate prediction of the biological outcome (inhibition, activation, no-response, mixed-response) in AR mutant-drug pairs compared to other machine learning approaches. Finally, the developed approach was used to make predictions of AR mutant response to the latest AR inhibitor darolutamide, which were then validated by in-vitro experiments.

HUME: large-scale detection of causal genetic factors of adverse drug reactions.

Motivation: Adverse drug reactions are one of the major factors that affect the wellbeing of patients and financial costs of healthcare systems. Genetic variations of patients have been shown to be a key factor in the occurrence and severity of many ADRs. However, the large number of confounding drugs and genetic biomarkers for each adverse reaction case demands a method that evaluates all potential genetic causes of ADRs simultaneously. Results: To address this challenge, we propose HUME, a multi-phase algorithm that recommends genetic factors for ADRs that are causally supported by the patient record data. HUME consists of the construction of a network from co-prevalence between significant genetic biomarkers and ADRs, a link score phase for predicting candidate relations based on the Adamic-Adar measure, and a causal refinement phase based on multiple hypothesis testing of quasi experimental designs for evaluating evidence and counter evidence of candidate relations in the patient records. Supplementary information: Supplementary data are available at Bioinformatics online.

Multi-treatment Effect Estimation from Biomedical Data.

Several biomedical applications contain multiple treatments from which we want to estimate the causal effect on a given outcome. Most existing Causal Inference methods, however, focus on single treatments. In this work, we propose a neural network that adopts a multi-task learning approach to estimate the effect of multiple treatments. We validated M3E2 in three synthetic benchmark datasets that mimic biomedical datasets. Our analysis showed that our method makes more accurate estimations than existing baselines.

Optimally discriminative subnetwork markers predict response to chemotherapy.

MOTIVATION: Molecular profiles of tumour samples have been widely and successfully used for classification problems. A number of algorithms have been proposed to predict classes of tumor samples based on expression profiles with relatively high performance. However, prediction of response to cancer treatment has proved to be more challenging and novel approaches with improved generalizability are still highly needed. Recent studies have clearly demonstrated the advantages of integrating protein-protein interaction (PPI) data with gene expression profiles for the development of subnetwork markers in classification problems. RESULTS: We describe a novel network-based classification algorithm (OptDis) using color coding technique to identify optimally discriminative subnetwork markers. Focusing on PPI networks, we apply our algorithm to drug response studies: we evaluate our algorithm using published cohorts of breast cancer patients treated with combination chemotherapy. We show that our OptDis method improves over previously published subnetwork methods and provides better and more stable performance compared with other subnetwork and single gene methods. We also show that our subnetwork method produces predictive markers that are more reproducible across independent cohorts and offer valuable insight into biological processes underlying response to therapy. AVAILABILITY: The implementation is available at: http://www.cs.sfu.ca/~pdao/personal/OptDis.html CONTACT: cenk@cs.sfu.ca; alapuk@prostatecentre.com; ccollins@prostatecentre.com.

High rate of de novo esophagitis 5 years after sleeve gastrectomy: a prospective multicenter study in Spain.

BACKGROUND: Major concerns years after the sleeve gastrectomy (SG) include weight regain, development of hiatal hernia (HH) and gastroesophageal reflux disease, with esophagitis and Barrett's esophagus (BE). Both problems could be related, and the incidence of asymptomatic patients is troubling. OBJECTIVE: To study the incidence of reflux symptoms, esophagitis, BE, HH, and asymptomatic pathology and their relationship with weight regain in patients 5 years after undergoing SG at different bariatric centers in Spain. SETTING: Public and private hospitals with bariatric surgery units. METHODS: Prospective, multicenter, nonrandomized study involving 13 Spanish hospitals with a cumulative experience of 4,500 patients having undergone the SG procedure and patients who had been subjected to the procedure at least 5 years previously along with preoperative gastroscopy. The clinical history, preoperative gastroscopy, and technical details of the SG were recorded. A specific clinical questionnaire was given that recorded the intake volume, perception of satiety, and gastroesophageal reflux (GER) symptoms. Gastroscopy, pH-metry, and manometry studies were carried out, and the data were analyzed statistically. The study has been authorized by the official Spanish ethics committee CEI/CEIm Hospital Universitario Gran Canaria Dr Negrín (code 2019-216-1). RESULTS: One hundred and five patients who underwent SG and who had with at least 5 years of follow-up were included. All procedures were performed laparoscopically. The mean age of patients was 51.1 years, and 70.5% were women. The mean characteristics of the SG procedure were a 37.2F probe, at 4.6 cm from the pylorus, and a crura closure was performed in 5 cases. There were no major complications (Clavien-Dindo grade >3) or deaths. The average preoperative body mass index was 46.3 kg/m2, the minimum reached was 20.6 kg/m2, whereas the average after 5 years was of 34.5 kg/m2. GER, HH, and esophagitis symptoms went from 17.1%, 28.6%, and 5.7%, respectively, before the SG to 76%, 30.5%, and 31.4%, respectively, 5 years after the procedure. Symptoms persisted over the years in 37.1% of cases and presented de novo in 52.8% of cases. Fifty-three percent of manometries (n = 27, total 51) and 64% of pH-metries (n = 32, total 53; DeMeester average score was 65) were pathologic 5 years after the procedure. Concerning gastroscopies, 5 years after the procedure, HH was found in 33 patients (30.5% of total) and esophagitis in 32 patients (31.4% of total). Eighty patients (76%) had GER symptoms, and 25 patients (24%) were asymptomatic. Only 1 patient (.9%) developed BE. CONCLUSIONS: Our study has confirmed a high rate of both persistent and de novo esophagitis and hiatal hernia, many of which were asymptomatic, 5 years after SG had been performed. Weight regain and a striking increase in gastric capacity are risk factors indicative of esophagitis, even when patients are asymptomatic. We consider a control gastroscopy and the preventive use of proton pump inhibitors necessary in these cases regardless of symptoms. We recommend that a control gastroscopy should be performed in all cases regardless of symptoms 5 years after SG. Further studies are needed to validate these recommendations.

Inferring cancer subnetwork markers using density-constrained biclustering.

MOTIVATION: Recent genomic studies have confirmed that cancer is of utmost phenotypical complexity, varying greatly in terms of subtypes and evolutionary stages. When classifying cancer tissue samples, subnetwork marker approaches have proven to be superior over single gene marker approaches, most importantly in cross-platform evaluation schemes. However, prior subnetwork-based approaches do not explicitly address the great phenotypical complexity of cancer. RESULTS: We explicitly address this and employ density-constrained biclustering to compute subnetwork markers, which reflect pathways being dysregulated in many, but not necessarily all samples under consideration. In breast cancer we achieve substantial improvements over all cross-platform applicable approaches when predicting TP53 mutation status in a well-established non-cross-platform setting. In colon cancer, we raise prediction accuracy in the most difficult instances from 87% to 93% for cancer versus non-cancer and from 83% to (astonishing) 92%, for with versus without liver metastasis, in well-established cross-platform evaluation schemes. AVAILABILITY: Software is available on request.

PSORTb 3.0: improved protein subcellular localization prediction with refined localization subcategories and predictive capabilities for all prokaryotes.

MOTIVATION: PSORTb has remained the most precise bacterial protein subcellular localization (SCL) predictor since it was first made available in 2003. However, the recall needs to be improved and no accurate SCL predictors yet make predictions for archaea, nor differentiate important localization subcategories, such as proteins targeted to a host cell or bacterial hyperstructures/organelles. Such improvements should preferably be encompassed in a freely available web-based predictor that can also be used as a standalone program. RESULTS: We developed PSORTb version 3.0 with improved recall, higher proteome-scale prediction coverage, and new refined localization subcategories. It is the first SCL predictor specifically geared for all prokaryotes, including archaea and bacteria with atypical membrane/cell wall topologies. It features an improved standalone program, with a new batch results delivery system complementing its web interface. We evaluated the most accurate SCL predictors using 5-fold cross validation plus we performed an independent proteomics analysis, showing that PSORTb 3.0 is the most accurate but can benefit from being complemented by Proteome Analyst predictions. AVAILABILITY: http://www.psort.org/psortb (download open source software or use the web interface). CONTACT: psort-mail@sfu.ca SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

ParKCa: Causal Inference with Partially Known Causes.

Methods for causal inference from observational data are an alternative for scenarios where collecting counterfactual data or realizing a randomized experiment is not possible. Our proposed method ParKCA combines the results of several causal inference methods to learn new causes in applications with some known causes and many potential causes. We validate ParKCA in two Genome-wide association studies, one real-world and one simulated dataset. Our results show that ParKCA can infer more causes than existing methods.

Moderate-Intensity Physical Activity, Music and Art Activities Preserved Cognitive Health in Older Adults: An Argument for Social Prescribing Solution.

Introduction: Rates of dementia are projected to increase over the coming years as global populations age. Without a treatment to slow the progression of dementia, many health policies are focusing on preventing dementia by slowing the rate of cognitive decline with age. However, it is unclear which lifestyle changes in old age meaningfully reduce the rate of cognitive decline associated with aging. Objectives: Use existing, multi-year longitudinal health data to determine if engagement in a variety of different lifestyle activities can slow the rate of cognitive decline as older adults age. Method: Data from the English Longitudinal Study of Aging was analyzed using a quasi-experimental, efficient matched-pair design inspired by the clinical trial methodology. Changes in short-term memory scores were assessed over a multi-year interval for groups who undertook one of 11 different lifestyle activities, compared to control groups matched across confounding socioeconomic and lifestyle factors. Results: Two factors, moderate-intensity physical activity and learning activities, resulted in significant positive impact on cognitive function. Conclusion: Our analysis brings cognitive benefit arguments in favor of two lifestyle activities, moderate-intensity physical activity and learning activities, while rejecting other factors advanced by the literature such as vigorous-intensity physical activity. Those findings justify and encourage the development of new lifestyle health programs by health authorities and bring forward the new health system solution, social prescribing.