Drug repurposing for COVID-19 based on an integrative meta-analysis of SARS-CoV-2 induced gene signature in human airway epithelium

Drug repurposing has the potential to bring existing de-risked drugs for effective intervention in an ongoing pandemic—COVID-19 that has infected over 131 million, with 2.8 million people succumbing to the illness globally (as of April 04, 2021). We have used a novel `gene signature’-based drug repositioning strategy by applying widely accepted gene ranking algorithms to prioritize the FDA approved or under trial drugs. We mined publically available RNA sequencing (RNA-Seq) data using CLC Genomics Workbench 20 (QIAGEN) and identified 283 differentially expressed genes (FDR<0.05, log2FC>1) after a meta-analysis of three independent studies which were based on severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) infection in primary human airway epithelial cells. Ingenuity Pathway Analysis (IPA) revealed that SARS-CoV-2 activated key canonical pathways and gene networks that intricately regulate general anti-viral as well as specific inflammatory pathways. Drug database, extracted from the Metacore and IPA, identified 15 drug targets (with information on COVID-19 pathogenesis) with 46 existing drugs as potential-novel candidates for repurposing for COVID-19 treatment. We found 35 novel drugs that inhibit targets (ALPL, CXCL8, and IL6) already in clinical trials for COVID-19. Also, we found 6 existing drugs against 4 potential anti-COVID-19 targets (CCL20, CSF3, CXCL1, CXCL10) that might have novel anti-COVID-19 indications. Finally, these drug targets were computationally prioritized based on gene ranking algorithms, which revealed CXCL10 as the common and strongest candidate with 2 existing drugs. Furthermore, the list of 283 SARS-CoV-2-associated proteins could be valuable not only as anti-COVID-19 targets but also useful for COVID-19 biomarker development.

TLGP: a flexible transfer learning algorithm for gene prioritization based on heterogeneous source domain

Background Gene prioritization (gene ranking) aims to obtain the centrality of genes, which is critical for cancer diagnosis and therapy since keys genes correspond to the biomarkers or targets of drugs. Great efforts have been devoted to the gene ranking problem by exploring the similarity between candidate and known disease-causing genes. However, when the number of disease-causing genes is limited, they are not applicable largely due to the low accuracy. Actually, the number of disease-causing genes for cancers, particularly for these rare cancers, are really limited. Therefore, there is a critical needed to design effective and efficient algorithms for gene ranking with limited prior disease-causing genes. Results In this study, we propose a transfer learning based algorithm for gene prioritization (called TLGP) in the cancer (target domain) without disease-causing genes by transferring knowledge from other cancers (source domain). The underlying assumption is that knowledge shared by similar cancers improves the accuracy of gene prioritization. Specifically, TLGP first quantifies the similarity between the target and source domain by calculating the affinity matrix for genes. Then, TLGP automatically learns a fusion network for the target cancer by fusing affinity matrix, pathogenic genes and genomic data of source cancers. Finally, genes in the target cancer are prioritized. The experimental results indicate that the learnt fusion network is more reliable than gene co-expression network, implying that transferring knowledge from other cancers improves the accuracy of network construction. Moreover, TLGP outperforms state-of-the-art approaches in terms of accuracy, improving at least 5%. Conclusion The proposed model and method provide an effective and efficient strategy for gene ranking by integrating genomic data from various cancers.

Drug Repurposing of potential drug targets for treatment of COVID-19

Drug repurposing has the potential to bring existing de-risked drugs for effective intervention in an ongoing pandemic-COVID-19 that has infected over 131 million, with 2.8 million people succumbing to the illness globally (as of April 04, 2021). We have used a novel `gene signature'-based drug repositioning strategy by applying widely accepted gene ranking algorithms to prioritize the FDA approved or under trial drugs. We mined publically available RNA sequencing (RNA-Seq) data using CLC Genomics Workbench 20 (QIAGEN) and identified 283 differentially expressed genes (FDR<0.05, log2FC>1) after a meta-analysis of three independent studies which were based on severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) infection in primary human airway epithelial cells. Ingenuity Pathway Analysis (IPA) revealed that SARS-CoV-2 activated key canonical pathways and gene networks that intricately regulate general anti-viral as well as specific inflammatory pathways. Drug database, extracted from the Metacore and IPA, identified 15 drug targets (with information on COVID-19 pathogenesis) with 46 existing drugs as potential-novel candidates for repurposing for COVID-19 treatment. We found 35 novel drugs that inhibit targets (ALPL, CXCL8, and IL6) already in clinical trials for COVID-19. Also, we found 6 existing drugs against 4 potential anti-COVID-19 targets (CCL20, CSF3, CXCL1, CXCL10) that might have novel anti-COVID-19 indications. Finally, these drug targets were computationally prioritized based on gene ranking algorithms, which revealed CXCL10 as the common and strongest candidate with 2 existing drugs. Furthermore, the list of 283 SARS-CoV-2-associated proteins could be valuable not only as anti-COVID-19 targets but also useful for COVID-19 biomarker development.

Exploring gene knockout strategies to identify potential drug targets using genome-scale metabolic models

Research on new cancer drugs is performed either through gene knockout studies or phenotypic screening of drugs in cancer cell-lines. Both of these approaches are costly and time-consuming. Computational framework, e.g., genome-scale metabolic models (GSMMs), could be a good alternative to find potential drug targets. The present study aims to investigate the applicability of gene knockout strategies to be used as the finding of drug targets using GSMMs. We performed single-gene knockout studies on existing GSMMs of the NCI-60 cell-lines obtained from 9 tissue types. The metabolic genes responsible for the growth of cancerous cells were identified and then ranked based on their cellular growth reduction. The possible growth reduction mechanisms, which matches with the gene knockout results, were described. Gene ranking was used to identify potential drug targets, which reduce the growth rate of cancer cells but not of the normal cells. The gene ranking results were also compared with existing shRNA screening data. The rank-correlation results for most of the cell-lines were not satisfactory for a single-gene knockout, but it played a significant role in deciding the activity of drug against cell proliferation, whereas multiple gene knockout analysis gave better correlation results. We validated our theoretical results experimentally and showed that the drugs mitotane and myxothiazol can inhibit the growth of at least four cell-lines of NCI-60 database.

A meta-analysis of the transcriptional landscape of end stage heart failure with reduced ejection fraction

During heart failure (HF) the myocardium undergoes ventricular remodeling, which is accompanied by changes in gene expression. Transcriptomic studies have contributed to fundamental knowledge of this remodeling. However many studies do not agree on the differentially expressed marker genes. We hypothesized that a comprehensive meta-analysis of available studies would help to address this incongruence. We curated and uniformly processed 14 public transcriptomic data sets consisting of left ventricular samples from 265 healthy and 642 failing hearts. We assessed their comparability by applying transfer learning based classifiers. We then combined all datasets to extract a consensus gene signature. Combining this consensus signature with biological prior knowledge, we estimated transcription factor activities with DoRothEA and pathway activities by footprint analysis with PROGENy. Although single studies reported highly dissimilar marker genes, transfer knowledge based approaches revealed that disease patterns are conserved across studies. We derived a gene ranking that reflects consistent molecular hallmarks of HF. The ranking is strongly enriched in fibrosis related gene sets and contains significant (all p-values &lt;0.05) footprints of numerous transcription factors including Nanog, Sox2, Pbx3, Mef2; pathways including Jak-Stat and miRNAs including MIR-206, MIR-514. We demonstrated the feasibility of combining transcriptional studies from different technologies, years and centers. We extracted a consensus gene signature as a valuable resource to understand common molecular patterns in HF. To our knowledge, this report represents the largest meta-analysis of transcriptional HF studies to date. Selected transcription factor (TF) activities derived via footprint analysis from the consensus gene ranking TF NES Size P-value SOX2 6.3 179 2.59E-10 MYNN −5.7 347 8.14E-09 MTA2 −5.3 29 9.1E-08 ZBTB7A 4.8 66 1.25E-06 BL3 −4.5 26 5.46E-06 ZEB2 −3.9 46 8.99E-05 NANOG 3.81 34 0.00047 PBX3 3.4 41 0.0005 MEF2C 3.1 55 0.001602 MEIS2 2.7 718369 0.00656 NES, normalized enrichment score; size, regulon size. Graphical summary Funding Acknowledgement Type of funding source: Foundation. Main funding source(s): Klaus Tschirer Stiftung

Genetic Modifiers at the Crossroads of Personalised Medicine for Haemoglobinopathies

Haemoglobinopathies are common monogenic disorders with diverse clinical manifestations, partly attributed to the influence of modifier genes. Recent years have seen enormous growth in the amount of genetic data, instigating the need for ranking methods to identify candidate genes with strong modifying effects. Here, we present the first evidence-based gene ranking metric (IthaScore) for haemoglobinopathy-specific phenotypes by utilising curated data in the IthaGenes database. IthaScore successfully reflects current knowledge for well-established disease modifiers, while it can be dynamically updated with emerging evidence. Protein–protein interaction (PPI) network analysis and functional enrichment analysis were employed to identify new potential disease modifiers and to evaluate the biological profiles of selected phenotypes. The most relevant gene ontology (GO) and pathway gene annotations for (a) haemoglobin (Hb) F levels/Hb F response to hydroxyurea included urea cycle, arginine metabolism and vascular endothelial growth factor receptor (VEGFR) signalling, (b) response to iron chelators included xenobiotic metabolism and glucuronidation, and (c) stroke included cytokine signalling and inflammatory reactions. Our findings demonstrate the capacity of IthaGenes, together with dynamic gene ranking, to expand knowledge on the genetic and molecular basis of phenotypic variation in haemoglobinopathies and to identify additional candidate genes to potentially inform and improve diagnosis, prognosis and therapeutic management.

Accurate Classification of Differential Expression Patterns in a Bayesian Framework With Robust Normalization for Multi-Group RNA-Seq Count Data

Empirical Bayes is a choice framework for differential expression (DE) analysis for multi-group RNA-seq count data. Its characteristic ability to compute posterior probabilities for predefined expression patterns allows users to assign the pattern with the highest value to the gene under consideration. However, current Bayesian methods such as baySeq and EBSeq can be improved, especially with respect to normalization. Two R packages (baySeq and EBSeq) with their default normalization settings and with other normalization methods (MRN and TCC) were compared using three-group simulation data and real count data. Our findings were as follows: (1) the Bayesian methods coupled with TCC normalization performed comparably or better than those with the default normalization settings under various simulation scenarios, (2) default DE pipelines provided in TCC that implements a generalized linear model framework was still superior to the Bayesian methods with TCC normalization when overall degree of DE was evaluated, and (3) baySeq with TCC was robust against different choices of possible expression patterns. In practice, we recommend using the default DE pipeline provided in TCC for obtaining overall gene ranking and then using the baySeq with TCC normalization for assigning the most plausible expression patterns to individual genes.

ClinPhen extracts and prioritizes patient phenotypes directly from medical records to accelerate genetic disease diagnosis

PurposeSevere genetic diseases affect 7 million births per year, worldwide. Diagnosing these diseases is necessary for optimal care, but it can involve the manual evaluation of hundreds of genetic variants per case, with many variants taking an hour to evaluate. Automatic gene-ranking approaches shorten this process by reporting which of the genes containing variants are most likely to be causing the patient’s symptoms. To use these tools, busy clinicians must manually encode patient phenotypes, which is a cumbersome and imprecise process. With 60 million patients expected to be sequenced in the next 7 years, a fast alternative to manual phenotype extraction from the clinical notes in patients’ medical records will become necessary.MethodsWe introduce ClinPhen: a fast, high-accuracy tool that automatically converts the clinical notes into a prioritized list of patient symptoms using HPO terms.ResultsClinPhen shows superior accuracy to existing phenotype extractors, and when paired with a gene-ranking tool it significantly improve the latter’s performance.ConclusionCompared to manual phenotype extraction, ClinPhen saves more than 5 hours per case in Mendelian diagnosis alone. Summing over millions of forthcoming cases whose medical notes await phenotype encoding, ClinPhen makes a substantial contribution towards ending all patients’ diagnostic odyssey.