Fig. 4 | Nature Communications

Fig. 4

From: A crowdsourced analysis to identify ab initio molecular signatures predictive of susceptibility to viral infection

Fig. 4

Overlap and pathway enrichment among predictors of symptoms. a Percent of team combinations showing statistically significant intersections of predictors at T0 and T24. Only teams with AUROC ≥ 0.5 or r ≥ 0 for subchallenges 2 and 3, respectively, were used for this analysis. The x-axis indicates the number of teams included in the combination. For example, the value 2 corresponds to pairwise overlaps, 3 corresponds to 3-way overlaps, etc. The y-axis indicates the percentage of team combinations with a statistically significant (p value < 0.05) predictor intersection. Point size indicates median intersection size of predictors among team combinations with significant predictor intersection; “X” indicates no significant predictor intersection. b Pathway enrichment among predictors of infection for each subchallenge (SC2 and SC3) at T0 and T24. The x-axis indicates subchallenge and each grid indicates timepoint. The y-axis indicates pathways enriched among predictors with a Benjamini−Hochberg-corrected Fisher’s exact test p value < 0.05. Point size represents the Fisher’s exact test enrichment −log10(p value). Point colors indicate whether the pathway was associated with symptoms (red) or lack thereof (blue). Pathways shared between both SC2 and SC3 at each timepoint are highlighted in gray. Pathways are ordered by the decreasing maxP test statistic as determined in Supplementary Figure 5. c GeneMANIA network of the union of predictors involved in the Heme metabolism pathway across timepoints (T0 and T24) and subchallenges (SC2 and SC3). Edges are inferred by GeneMANIA51 corresponding to coexpression (gray), physical interactions (orange), and genetic interactions (purple) among genes. Node size corresponds to the number of teams that selected the predictor

Back to article page