includes DNA variants such as singlenucleotide polymorphisms (SNPs), variability of short sequence repeats, haplotype, etc. Outcomes: clinical diagnosis of SARS or associated syndromes; evidence of infection; atypical manifestation of illness; measurements of clinical outcomes (mortality, recurrence, severity, quality of life, etc.). Main outcome measure: odds ratio. Ancestry and sub-phenotypes were pre-defined as possible stratification parameters in data synthesis. Study design: all study types showing effect size and significance were eligible (RCTs, cohort, case-control and cross-sectional studies). Type of article: original peer-reviewed articles reporting quantitative results of the association between the CoVs-related phenotypes and any germline genomic variant. Exclusion criteria: editorials and opinion papers were not considered. Pertinent narrative and systematic reviews were tagged in the screening phase and inspected for relevant references. Preprint articles that had not peer-reviewed were not eligible, though were considered for discussion. Strategy for literature search, data collection and data synthesis The following databases were inspected: PubMed and Scopus for indexed published articles; medRxive and bioRxive for pre-print articles. The search strategy was developed by two expert reviewers following an iterative process to ensure sensitivity. Two domains (“CoV” and “Genetics”) were defined combining MeSH terms and keywords using the Boolean operator “or”; the two domains were combined with “and”. Keywords for the dimension “genetics” were searched in the title field only. Several test runs showed that sensitivity was not affected, despite the remarkable reduction in the number of records retrieved [data not shown]. After interim results, a domain was added (“CoV-associated Biomarkers”) which targeted specific genes and genomic biomarkers found during the preliminary inclusion phase; the domain was combined with the “CoV” dimension with “and”; the resulting records were checked for duplicates and screened, then added to the final list of included articles. Search strings are reported in Supplementary Table S1. Records were filtered for language (English) and date of publication (Jan 2003 to Apr 2020). The publication year field was restricted to the period from 2003 onwards, to include all articles published after the SARS outbreak dated 2003. The distribution of relevant publications indexed in PubMed confirmed a sharp rise since 2003 and a massive increase in 2020 (not shown). Given the rapid evolution of the field, the literature search was replicated on June 17, in order to retrieve the Di Maria et al. Human Genomics (2020) 14:30 Page 3 of 19 latest articles and published versions of pre-print documents. To this purpose, PubMed and Scopus were interrogated. Each identified study was indexed and manually abstracted. A total of 37 fields were pre-set for data extraction. Findings for all individual genetic variants under investigation were recorded, whether significantly associated or not. Some studies provided more than one data point (corresponding to one variant—one outcome measure entry) and were consequently represented with more than one record. The consistency of results between studies was analysed by gene and by the individual genomic biomarker. The study protocol did not envisage clinical recommendations after data synthesis. Owing to the need for a rapid appraisal, the following modifications of the standard protocol for systematic reviews were applied, according to the current recommendations for rapid reviews [25, 26]. Grey literature was not systematically searched. Only articles in English were included in the screening phase and considered as eligible. One reviewer completed the screening phase according to an expedited PICO-based method previously described [27] and applying the rule of thumb “if in doubt keep it in” [24]. A second reviewer screened a sample of records. Eligible articles were inspected in-full by one reviewer and verified on title and abstract by a second independent reviewer. In case of doubts, the two reviewers discussed the accordance with the inclusion criteria. Data abstraction was drafted by one reviewer and verified independently by a second reviewer. Data synthesis was drafted by one reviewer and verified by all authors. Raw data were not requested. A formal assessment of the risk of bias was omitted. We relied on peer-review to ensure that included studies were methodologically sound, including proper and reliable statistical analysis (e.g. threshold for significance, correction for multiple testing, etc.). Therefore, unrefereed preprints were not included. It is noteworthy that most genetic association studies are based on crosssectional or case-control design and as such are intrinsically affected by a high risk of bias. Results Systematic search The systematic literature search in PubMed, Scopus, medRxive and bioRxive provided 1567 unique records and additional 362 records found in the last update (June 17, 2020). After the screening phase and full-text inspection, 32 articles fulfilled the criteria and were eligible for data abstraction. The detailed flow of literature assessment according to the PRISMA statement is depicted in Supplementary Figure 1. Considering all variants tested as a predictor and the different