COMBATdb: a database for the COVID-19 Multi-Omics Blood ATlas

Abstract

Advances in our understanding of the nature of the immune response to SARS-CoV-2 infection, and how this varies within and between individuals, is important in efforts to develop targeted therapies and precision medicine approaches. Here we present a database for the COvid-19 Multi-omics Blood ATlas (COMBAT) project, COMBATdb (https://db.combat.ox.ac.uk). This enables exploration of multi-modal datasets arising from profiling of patients with different severities of illness admitted to hospital in the first phase of the pandemic in the UK prior to vaccination, compared with community cases, healthy controls, and patients with all-cause sepsis and influenza. These data include whole blood transcriptomics, plasma proteomics, epigenomics, single-cell multi-omics, immune repertoire sequencing, flow and mass cytometry, and cohort metadata. COMBATdb provides access to the processed data in a well-defined framework of samples, cell types and genes/proteins that allows exploration across the assayed modalities, with functionality including browse, search, download, calculation and visualisation via shiny apps. This advances the ability of users to leverage COMBAT datasets to understand the pathogenesis of COVID-19, and the nature of specific and shared features with other infectious diseases.

Figure 1.

Schematic summarising the main modules of COMBATdb.

Figure 2.

Examples of COMBATdb pages. (A–E) Lists of browsable (A) assay modalities (B) sources (comparator groups) (C) participants (D) genes (E) cell types. (F) Landing page for multiomics data single search mode and the results. (G) Example of results for samples from Flow Cytometry: FACS modality. (H) Landing page for compare functionality and the resulting enriched Reactome pathways for Proteomics: timsTOF modality. (I) List of browsable features and components for various data types on the landing page of tensor decomposition modality.

Figure 3.

Examples of analysis modules in shiny apps. (A–E) Plasma proteomics: timsTOF modality (A) options on the visualisation module (B, C) analysis generated from ‘One priority sample at maximum severity per individual’ (B) principal component analysis (PCA) plot coloured by source comparator groups (C) PCA loadings plot labelled by protein names. (D) Volcano plot based on the results from differential abundance analysis between COVID-19 (critical) and COVID-19 (mild). (E) Boxplot for the protein expression levels of a given protein (LRG1) grouped by source comparator groups. (F) Boxplot of LRG1 from ‘Bulk RNA-Seq modality’. (G) Boxplot of LRG1 from ‘CITE-Seq: GEX pseudobulk modality’ with cell group resolution ‘cell type’ and cell cluster name ‘MNP’ (mononuclear phagocytes).

Figure 4.

Integrative analysis using tensor decomposition. (A) Options within the visualisation module of the tensor decomposition modality. (B–E) Tensor decomposition shiny app showing (B) boxplot of loading scores of samples grouped by comparator group sources (C) barplot of loading scores for nine cell types for expression (D) barplot of loading scores for timsTOF proteins (E) barplot of loading scores of genes.