Interaction databases

The biocuration process was supported by interaction and pathway databases storing structured, annotated and curated information about COVID‐19 virus–host interactions. The IMEx Consortium (Meldal et al, 2019) dataset (Perfetto et al, 2020) contains curated Coronaviridae‐related interaction data from reviewed manuscripts and preprints, resulting in a dataset of roughly 7,300 interactions extracted from over 250 publications, including data from SARS‐CoV‐2, SARS, CoV, and other strains of Coronaviridae. The dataset is updated with every release of IMEx data and is open access (https://www.ebi.ac.uk/intact/resources/datasets#coronavirus). The SIGNOR 2.0 (Licata et al, 2020) dataset contains manually annotated and validated signalling interactions related to the host–virus interaction, including cellular pathways modulated during SARS‐CoV‐2 infection. The dataset was constructed from the literature on causal interactions between SARS‐CoV‐2, SARS‐COV‐1, MERS proteins and the human host and is openly available (https://signor.uniroma2.it/covid/). The Elsevier Pathway Collection (Daraselia et al, 2004; Nesterova et al, 2020) COVID‐19 dataset comprises manually reconstructed and annotated pathway diagrams. Statements about molecular interactions are extracted into a knowledge graph by a dedicated text mining technology adapted for extracting facts about viral proteins and viruses from the literature. These interactions were filtered for experimental evidence, used for pathway reconstruction and made openly available (http://dx.doi.org/10.17632/d55xn2c8mw.1). Information from OmniPath (Türei et al, 2021) on existing interactions gathered from pathway and interaction databases was used in a programmatic way to suggest cell‐specific interactions and cell–cell interactions specific to immune reactions.