Bowtie Assembly Parameters

Question

[biolab]

Hi, everyone, I am trying Bowtie for reference based assembly. My draft genome dataset is not from a model organism, but a related species. This data contains ~20 million 88 bp illumina reads. Because it's not resequencing of reference genome, I need to use a relatively loose parameters. Actually I only focus on genes without caring about intergenic regions. A CDS dataset of model organism can be used as a reference in this assembly. What Bowtie parameters you suggest to use? Thank you very much.

[Devon Ryan]

Bowtie is an aligner, not an assembler. Why don't you just use a splice-aware aligner, like tophat, or just try de novo assembly (e.g., with trinity)?

[r.follador]

Can you actually use trinity for genome assembly? I thought it's specifically for transcript assembly.

[Devon Ryan]

Since biolab mentioned aligning to CDS, I'm assuming that he/she is actually doing RNAseq (though, upon rereading, this may well be incorrect!). Otherwise, yes, I think you're correct.

Answer 1

[Adrian Pelin]

I concur with the first comment, specifically I suggest an assembler. Your data is 88bp illumina, is it paired end? That will improve your assembly statistics. Here are my suggestions:

1) Try SPAdes or Velvet, read here about how they differ. 2) You can map your resulting contigs to your close species. This is better, since contigs are much much longer than reads, and you can virtually see which parts of the genome remain homologous.

[biolab]

thanks for ur suggestions.

[c.v.oflynn]

It depends on how similar your reference is. bwa is quite good for losely mapping, or "last" given you have enough computational resource. Then just parse the outputs incorporating snps/indels.

[biolab]

thanks for your suggestions, all helpful!