Hi, I am wondering if anyone has used short read data to make MLST profiles. I imagine it would be more time-efficient and maybe even more accurate to map instead of making a de novo assembly. I've only seen a few hints at it in a blog post from Pathogenomics. These include using ICORN and SRST. I think that the other methods on the post use de novo assembly, which I am trying to avoid. ICORN seems like it could be rather involved, but I'm not sure yet, and SRST reportedly doesn't work with newer versions of Samtools.
So has anyone actually hammered out a set of scripts to map reads against an MLST database and determine a profile?
One difficulty that I can imagine is getting the right sequence on the ends of each MLST fragment. Maybe there are more difficulties.
I found that SRST has been updated since that blog post and am trying it out now, but any other choices would be helpful. Thank you.
Apart from the totally unhelpful answer below, this remains unanswered. I am very hopeful that someone has the right solution.
I have worked with SRST a little because it is ideally the right answer, but the software has a few bugs. So far I have altered the main executable in the following ways:
However, it is choking on samtools.pl pileup2fq. I'll see if I can fix it and will write down how to fix the software when/if I can. I believe that I will have change samtools pileup to samtools mpileup somewhere, but I am not sure yet.