Bowtie to align reads to single chromosome or region?

[jjw14]

Question for the Bowtie experts out there:

If I was to generate a bowtie index from a single chromosome, rather than the full genome, and align my reads to this "chromosome-index", would you expect any sort of bias/error in the alignment results? This would be for 42 bp, single end reads from size-selected genomic DNA.

I expect that if Bowtie's options are set to only align unique reads with no mismatches, there shouldn't be a difference between whole-genome and single chromosome index alignments. However, if mismatches were allowed, I would expect that more reads would align to the single chromosome index. Does this sound correct?

I will test it out and post my findings, but if someone already knows the answer, I'd really appreciate hearing more about it.

Thanks in advance,
jjw

[rwenang]

Hi, just my 2cents:

"I expect that if Bowtie's options are set to only align unique reads with no mismatches, there shouldn't be a difference between whole-genome and single chromosome index alignments"

I think it will differ quite significantly. A single chromosome mapping will have more chance of finding unique reads compared to whole genome. It is highly possible that there are duplications of some 40-mer on other chromosomes, which in the case of single chromosome mapping will be regarded as unique reads, but in the case of whole genome mapping will be regarded as multiple matches.

"However, if mismatches were allowed, I would expect that more reads would align to the single chromosome index."

I also thinks that given allowed mismatches, more reads will align to whole genome mapping rather than the single chromosome mapping.

The principle is that the longer sequence you use, the higher the probability you will get 40-mer duplication.