Fig. 1

Prediction and characterization of candidate structured RNAs. a Workflow of candidate structured RNA prediction. Genes in HMP2 [9] are annotated with Prodigal [10]. Homologous intergenic regions are identified using HS-BLASTN [8]. Conserved regions are clustered and scored using RNAphylo [2], R-scape [11], and RNAcode [12]. These putative structured RNAs are filtered to exclude those already found in Rfam 14.3 [13]. Candidate structured RNAs are searched against HMP2 using cmsearch [14] to ensure strong, unique hits were observed for each proposed structure. To triple check candidate structured RNAs are in non-coding regions, we search regions against the nr database using BLASTx [15]. This results in 3161 candidate structured RNAs, in addition to 2022 candidate structured RNAs with regions overlapping the nr database. b Histogram displaying the ten protein domains that are most commonly found in the genomic neighborhoods (within 5 kilobases) of the 3161 candidate structured RNAs. c Venn diagram displaying the taxonomic distributions of the 3161 candidate structured RNAs at the domain level based on One Codex [16] analysis of contigs on which they are found