Isolation and sequencing of AGO-bound RNAs reveals characteristics of mammalian stem-loop processing in vivo

Ian M. Silverman; Sager J. Gosai; Nicholas Vrettos; Shawn W. Foley; Nathan D. Berkowitz; Zissimos Mourelatos; Brian D. Gregory

Identical reads were collapsed but clone information was retained to reduce computational time. Trimmed reads were sequentially mapped to miRBase (v20) and RefSeq annotated spliced transcript models (hg19 or mm10, downloaded on 06082015), with a two-stage alignment strategy with Bowtie2 and EMBOSS-WATER. First, Bowtie2 was used to map the 5' regions of reads to either miRbase (v20) primary miRNA sequences with a 50 nt extension on the 3' end or RefSeq annotated spliced transcript models. For pre-miRNA-seq, the first 35 nt were used in the initial alignment step, whereas 18 nt were used for miRNA-and smRNA-seq. Following Bowtie2 alignment, reads were extended by local . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/294488 doi: bioRxiv preprint alignment with EMBOSS-WATER (with parameters: -gapopen 10.0 -gapextend 0.5). Multimapped reads were partially resolved by selecting the longest, highest scoring alignments. Mismatches detected at the 3' ends of reads were considered as non-templated additions and analyzed separately. We filtered unmapped reads from pre-miRNA-seq for known rRNA, snoRNA, snRNA, tRNA, mitochondrial transcripts, and repeat-masked sequences before mapping to RefSeq.