Gene set proximity analysis: expanding gene set enrichment analysis through learned geometric embeddings, with drug-repurposing applications in COVID-19.

MOTIVATION: Gene set analysis methods rely on knowledge-based representations of genetic interactions in the form of both gene set collections and protein-protein interaction (PPI) networks. However, explicit representations of genetic interactions often fail to capture complex interdependencies among genes, limiting the analytic power of such methods. RESULTS: We propose an extension of gene set enrichment analysis to a latent embedding space reflecting PPI network topology, called gene set proximity analysis (GSPA). Compared with existing methods, GSPA provides improved ability to identify disease-associated pathways in disease-matched gene expression datasets, while improving reproducibility of enrichment statistics for similar gene sets. GSPA is statistically straightforward, reducing to a version of traditional gene set enrichment analysis through a single user-defined parameter. We apply our method to identify novel drug associations with SARS-CoV-2 viral entry. Finally, we validate our drug association predictions through retrospective clinical analysis of claims data from 8 million patients, supporting a role for gabapentin as a risk factor and metformin as a protective factor for severe COVID-19. AVAILABILITY AND IMPLEMENTATION: GSPA is available for download as a command-line Python package at https://github.com/henrycousins/gspa. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Combinatorial Analysis of Phenotypic and Clinical Risk Factors Associated With Hospitalized COVID-19 Patients.

Characterization of the risk factors associated with variability in the clinical outcomes of COVID-19 is important. Our previous study using genomic data identified a potential role of calcium and lipid homeostasis in severe COVID-19. This study aimed to identify similar combinations of features (disease signatures) associated with severe disease in a separate patient population with purely clinical and phenotypic data. The PrecisionLife combinatorial analytics platform was used to analyze features derived from de-identified health records in the UnitedHealth Group COVID-19 Data Suite. The platform identified and analyzed 836 disease signatures in two cohorts associated with an increased risk of COVID-19 hospitalization. Cohort 1 was formed of cases hospitalized with COVID-19 and a set of controls who developed mild symptoms. Cohort 2 included Cohort 1 individuals for whom additional laboratory test data was available. We found several disease signatures where lower levels of lipids were found co-occurring with lower levels of serum calcium and leukocytes. Many of the low lipid signatures were independent of statin use and 50% of cases with hypocalcemia signatures were reported with vitamin D deficiency. These signatures may be attributed to similar mechanisms linking calcium and lipid signaling where changes in cellular lipid levels during inflammation and infection affect calcium signaling in host cells. This study and our previous genomics analysis demonstrate that combinatorial analysis can identify disease signatures associated with the risk of developing severe COVID-19 separately from genomic or clinical data in different populations. Both studies suggest associations between calcium and lipid signaling in severe COVID-19.

Determining Cost-Optimal Next-Generation Sequencing Panels for Rare Disease and Pharmacogenomics Testing.

BACKGROUND: Multi-gene panel sequencing using next-generation sequencing (NGS) methods is a key tool for genomic medicine. However, with an estimated 140 000 genomic tests available, current system inefficiencies result in high genetic-testing costs. Reduced testing costs are needed to expand the availability of genomic medicine. One solution to improve efficiency and lower costs is to calculate the most cost-effective set of panels for a typical pattern of test requests. METHODS: We compiled rare diseases, associated genes, point prevalence, and test-order frequencies from a representative laboratory. We then modeled the costs of the relevant steps in the NGS process in detail. Using a simulated annealing-based optimization procedure, we determined panel sets that were more cost-optimal than whole exome sequencing (WES) or clinical exome sequencing (CES). Finally, we repeated this methodology to cost-optimize pharmacogenomics (PGx) testing. RESULTS: For rare disease testing, we show that an optimal choice of 4-6 panels, uniquely covering genes that comprise 95% of the total prevalence of monogenic diseases, saves $257-304 per sample compared with WES, and $66-135 per sample compared with CES. For PGx, we show that the optimal multipanel solution saves $6-7 (27%-40%) over a single panel covering all relevant gene-drug associations. CONCLUSIONS: Laboratories can reduce costs using the proposed method to obtain and run a cost-optimal set of panels for specific test requests. In addition, payers can use this method to inform reimbursement policy.