Reactome COVID-19: the literature curation strategy
<p>ABSTRACT</p> <p>In response to the deluge of COVID-19-related publications Reactome developed a computational triaging strategy to review and identify publications appropriate for manual curation (66,100 SARS-Cov-2 articles on PUBMED, tallied on 30/October/2020 https://www.ncbi.nlm.nih.gov/research/coronavirus/; Chen et al., 2020). </p> <p>The literature triaging approach consisted of 4 main steps: 1) Literature screening; 2) Literature selection; 3) Reference tagging; and 4) Reference database construction. Two primary reference databases were downloaded and automatically text-mined: CDC COVID-19 downloadable database and bioRxiv database. Other collections of SARS-Cov-2 literature were manually screened with a focus on molecular interactions. These included: a Zotero Library built and updated by members of COVID-19 Disease Map (Ostaszewski et al., 2020); CORD-19 (Wang et al., 2020); LitCOVID (Chen et al., 2020); Johns Hopkins literature summary; Cell Press Coronavirus Resource Hub; Nature Coronavirus and COVID-19 updates; and Science’s Latest Coronavirus research. </p> <p>About 5% of the articles made it through reference screening focused on Reactome SARS-CoV-2 map construction to the literature selection step. If relevance was confirmed, the reference was then tagged in step 3. SARS-CoV-2 selected references were tagged regarding multiple features: (i) type of publication (e.g., article, review, pre-print, comment); (ii) Virus and host species (e.g., SARS-CoV-2, SARS-CoV-1; MERS, ACE2); (iii) Entity (specific molecules studied); (iv) Methods (e.g., Cryo-EM, ELISA, IC50); (v) Cell line and/or Tissue (e.g., vero-E6, lung tissue); (vi) subcellular localization (e.g., plasma membrane, ER); (vii) Molecular event (e.g., virus cycle step, pathway, host response); and (viii) Phenotype (e.g., immune, coagulation). Features i, ii, iii, vii and viii were mandatory. This Reference Database is stored in a shared spreadsheet, in which Reactome team members can edit and refine the Library (e.g., inclusion of tags). </p> <p> As Reactome is an evidence-based database built on reliable experimental published data, the process of SARS-CoV-2 reference selection is stringent and prioritizes peer-reviewed references. Nevertheless, the final decision on the reliability of the scientific evidence to support a molecular interaction was made by Reactome curators. In this case the focused literature triaging provides curators with a deeply researched trove of articles.</p> <p> The Reactome strategy of first creating an individual map for SARS-CoV-1 supported and guided the construction of a refined SARS-CoV-2 map. The SARS-CoV-2 Reactome map is an ongoing task, built on the foundation of a strong literature curation strategy. Together the literature triage and curatorial groups have built open-source COVID-19 viral infection pathways incorporating rapid scientific development and literature availability.</p> <p> </p> <p>References.</p> <p> </p> <p>Chen, Q., Allot A., Lu Z. Keep up with the latest coronavirus research. Nature 579, 193 (2020). doi: 10.1038/d41586-020-00694-1 </p> <p> </p> <p>Ostaszewski, M., Mazein, A., Gillespie, M.E. et al. COVID-19 Disease Map, building a computational repository of SARS-CoV-2 virus-host interaction mechanisms. Sci Data 7, 136 (2020).<a href="https://doi.org/10.1038/s41597-020-0477-8"> https://doi.org/10.1038/s41597-020-0477-8</a></p> <p> </p> <p>Wang, L., Lo K, Chandrasekhar, Y., et al. CORD-19: The Covid-19 Open Research Dataset. Preprint. ArXiv. 2020;arXiv:2004.10706v2. Published 2020 Apr 22.CORD-19.<a href="https://covidsearch.sinequa.com/app/covid-search/#/home"> https://covidsearch.sinequa.com/app/covid-search/#/home</a></p> <p> </p> <p>Reference Databases: CDC Covid-19 Database (<a href="https://www.cdc.gov/library/researchguides/2019novelcoronavirus/researcharticles.html">https://www.cdc.gov/library/researchguides/2019novelcoronavirus/researcharticles.html</a>); bioRxiv database (<a href="https://www.biorxiv.org/about-biorxiv">https://www.biorxiv.org/about-biorxiv</a>); Johns Hopkins literature summary (<a href="https://ncrc.jhsph.edu/topics/">https://ncrc.jhsph.edu/topics/</a>); Cell Press Coronavirus Resource Hub (<a href="https://www.cell.com/COVID-19">https://www.cell.com/COVID-19</a>); Nature Coronavirus and COVID-19 updates (<a href="https://www.nature.com/collections/aijdgieecb">https://www.nature.com/collections/aijdgieecb</a>); Science’s Latest Coronavirus research (<a href="https://www.sciencemag.org/collections/coronavirus?IntCmp=coronavirussiderail-128">https://www.sciencemag.org/collections/coronavirus?IntCmp=coronavirussiderail-128</a>).</p> <p> </p>
ShareScore
32/100
Overall dataset sharing score
Score breakdown
These five areas show where the dataset supports — or may limit — practical reuse.
- Stewardship
- 4
- Harmonization
- 4
- Access
- 16
- Reuse readiness
- 8
- Engagement
- 0