Best Bwa Parameters For Mirna Alignment

Question

[Doctoroots]

im analyzing a miRNA run from illumina genome analyzer. i received the data with the adapter already trimmed off, and im now at the point of aligning my reads against a reference.
when i try and align the reads using bwa's default params:

bwa aln ref.fasta read.fasta > read.sai

im getting <10% alignment (~90% of the reads are not getting aligned).

when i tried aligning using novoalign (that has the -s parameter to allow trimming) around 50% of the reads were aligned. this leads me to believe there is a parameter set in bwa that is more suitable for short RNA (including miRNA) alignment.

thoughts?

Answer 1

[Michael]

Some details worth considering:

How long are your reads? How long are the miRNAs you are looking for?

I have also seen on my data that bwa reported fewer alignments than some other aligners (even blat and lastz) given the standard parameters.

There are some parameters worth a check:

-n NUM    max #diff (int) or missing prob under 0.02 err rate (float) [0.04]

Try increasing -n

-o INT    maximum number or fraction of gap opens [1]

Try increasing to 2 or 3

-l INT    seed length [32]

the default 32 seems a little large to me especially if your query is short, try reducing towards 8 if that is possible

I would also try bowtie, bfast and maybe some others. For all the seed length parameter is important to increase sensitivity.