This is a test version of Biostars. For the public version, visit https://www.biostars.org.
STAR for bacterial genome

Hi Biostars,

Is it fine to use STAR for bacterial data (no splicing)? Any comments/suggestions are highly appreciated.

Thanks

star prokaryots
• 8,396 views • link
updated by Constanza • Constanza • written by grant.hovhannisyan grant.hovhannisyan

1 answer

It should work but you need to specify --alignIntronMax 1 to force STAR to avoid splice alignments.

Also during the genome index generation step you should put --genomeSAindexNbases to min(14, log2(GenomeLength)/2 - 1) if your bacterial genome of interest is relatively small. GenomeLength is in base pairs

• 0 views • link
written by Nicolas Rosewick Nicolas Rosewick

Thanks, very helpful. Are you aware of any comparison (paper or post) between STAR and lets say bowtie2 or bwa in case of bacterial data?

• 0 views • link
written by grant.hovhannisyan grant.hovhannisyan

No don't know any benchmarking for bacterial genomes

• 0 views • link
written by Nicolas Rosewick Nicolas Rosewick

I've made quite a few comparisons, but haven't published them yet.

Briefly, bacterial genomes are quite easy to map to - so uniquely mapped reads don't differ much between bwa/bowtie2/STAR/etc. However STAR/hisat2 are a lot better in reporting multimappers, which could be interesting for highly repetitive genomes like Nesseria etc.

• 0 views • link
written by predeus predeus

Do you have any experience with EDGE-pro, since that program should be made specifically for bacterial RNA-seq?

• 0 views • link
written by Line-hede-knudsen • Line-hede-knudsen

EDGE-pro is using bowite2, reporting up to 10 multimappers.

• 0 views • link
written by predeus predeus

I think the formula might be over-estimating the value a bit. The number min(14, log2(GenomeLength)/2 - 1) for something like Ecoli/Salmonella should be approximately 10, yet sometimes I get segfaults with this setting. Changing it to 8 fixes everything.

• 0 views • link
written by predeus predeus

Hi, Nicolas. What do you understand by "relatively small genomes"? I am dealing with data that ranges from 1,02 Mb to 9,56 Mb. Biologically, I've read publications that indicate 2 Mb as a common threshold but I'm guessing that the impact of the parameter won't attend to considerations of that nature when computing. Thanks in advance.

• 0 views • link
written by Constanza • Constanza

Log in to answer this question.

More posts like this