Entering edit mode
2.2 years ago
bright602
▴
50
Hi there, I just want to ask whether it is problematic to use bowtie2 for RNAseq mapping and refrain from using it? Thanks a lot for your insights!
Technically, it's perfectly fine if you use bowtie2+RSEM (which doesn't use genome alignment).
However, it is very very slow.
I used it in the RSEM pipeline where I also have STAR as an option. Just want to confirm using bowtie2 in RSEM pipeline. Thanks so much!
You're welcome! Yup, like I said above, bowtie2+RSEM works great aside from speed. RSEM (whether using STAR, like ENCODE does, or bowtie2) is also much more accurate than doing featureCounts on genome alignments (there are many published benchmarks on this).
Yes, but I was assuming genome alignments. Sure, you can align to the spliced transcriptome. I cannot back this up with a personal benchmark but I kind of feel safer to have a genome (and be it a decoy) in the mapping. I personally use salmon with the genome-decoy to capture gDNA contaminants and other spurious transcriptomic alignments.
Agreed, and also, there are plenty of publications using it and its being mainteined very frequently, with the last update beign from Jan 16, 2022. There are however some alternatives that are more recent and come as part of the "new tuxedo", like hisat2