Table 1. Ten molecular signatures from genomics data used for predicting novel RLR pathway components.
| Group | Molecular signature | Data set description | Type a | References | Number of genes b | Likelihood ratio score b , c |
|---|---|---|---|---|---|---|
| Virus-based | Positive selection in primates | Rapidly evolving genes in the primates lineage, detected using maximum likelihood analysis of nucleotide alignments | d | [23] | 926 | 1.7 |
| PPI with viruses | Virus-interacting human proteins extracted from PPI databases | c | [28] | 2,587 | 4.2 | |
| Viral miRNA target | Likelihood scores of targeting of human transcripts by viral miRNAs, based on predicted target sites | c | [30] | 6,761 | 1.3 | |
| Differential expression upon infection | Genes showing differential expression in lung epithelial cells infected with four respiratory viruses | c | see Methods | 1,680 | 3.5 | |
| Antiviral host factor | Meta analysis of genes with antiviral activity from seven RNAi screens studying a variety of viruses | c | see S3 Table | 173 | 2.1 | |
| Pathway-based | Co-expression with RLR pathway | Weighted co-expression with known RLR genes across >450 human gene expression studies | c | [32] | 4,149 | 2.4 |
| RLR pathway protein domain | Proteins containing one of the 25 domains that are significantly enriched in known RLR proteins | c | [66] | 711 | 8.9 | |
| Innate antiviral TF binding motifs | Genes with IRF, AP–1, NFκB, or STAT TF binding motifs in their promoters, based on conservation across 29 mammals | c | [33] | 4,508 | 2.3 | |
| NFκB activation mediator | Hits from a genome-wide siRNA screen of Epstein-Barr virus-induced NFκB activation | d | [34] | 154 | 19.8 | |
| RLR pathway PPI | Proteins that interact with known RLR proteins, calculated from PPI data | c | [35] | 1,750 | 4.3 | |
| Integration | RLR score | Bayesian integration of the 10 molecular signatures to predict novel RLR pathway components | ||||
a Data used directly (d) or as basis for further calculation (c)
b Combination of all bins with positive likelihood ratio scores per feature, derived from Fig 1A
c RLR genes versus non-RLR genes: P(Di | RLR genes) / P(Di | non-RLR genes), see Methods.
Note that, to avoid circularity, the predictive ability of the co-expression, protein domain and RLR pathway PPI data sets was assessed using the set of TLR, CLR, NLR, cytDNA genes instead of the RLR genes (see Methods).