PerM is an ultra-fast and sensitive SOLiD reads mapping tool

[KevinLam]

spotted this in solid software tools page.

PerM perm PerM is an ultra-fast and sensitive SOLiD reads mapping tool. It maps reads in csfasta (w/wo qual), fasta or fastq format to the genome or transcriptoms in fasta. The output could be the SAM format. PerM can be full sensitive to four mismatches and highly sensitive to more mismatches.

http://code.google.com/p/perm/

anyone using it?

[Kevin_Johnson]

I've used the PerM software. I prefer Bowtie for Illumina, but PerM is by far and away the best software for SOLiD reads. The authors have constantly improved upon it and actually were receptive to some of my suggestions.

One of the authors even gave me a preliminary version of their "ComB" software which calls SNPs, consensus genomes, and does expression in color space. Def the way to go for SOLID.

[nilshomer]

I've used the PerM software. I prefer Bowtie for Illumina, but PerM is by far and away the best software for SOLiD reads. The authors have constantly improved upon it and actually were receptive to some of my suggestions.

One of the authors even gave me a preliminary version of their "ComB" software which calls SNPs, consensus genomes, and does expression in color space. Def the way to go for SOLID.

No support for gaps (indels), which can lead to false-mappings as well as ignoring a very important source of variation. This pertains to bowtie and PerM. I think the inclusion of gapped alignment would definitely put it in contention.

[Kevin_Johnson]

Actually, I just talked to a PerM author and it seems that supporting gaps actually leads to false mappings! Ironic isn't it. It's because of the biology; gaps are very rare in comparison to snps and sequencing errors, so most gaps on short reads are just false positives, ie. allowing a gap makes the rest of the sequence match well by chance!

Anyway, I have been using their tools a lot, between comb and the ClipR tool, they have really accurate snp calling and deletion finding. However, it seems like the new tools are buggy.

Anyway, one thing I dont understand about bfast, is I keep getting this error:
In function "BfastIndexValidateInputs": Fatal Error[OutOfRange]. Variable/Value: indexNumber.
Message: Command line argument.
***** Exiting due to errors *****

I'm using the same settings as I used before for shorter reads without any problems? what should i change?

[nilshomer]

Actually, I just talked to a PerM author and it seems that supporting gaps actually leads to false mappings! Ironic isn't it. It's because of the biology; gaps are very rare in comparison to snps and sequencing errors, so most gaps on short reads are just false positives, ie. allowing a gap makes the rest of the sequence match well by chance!

Anyway, I have been using their tools a lot, between comb and the ClipR tool, they have really accurate snp calling and deletion finding. However, it seems like the new tools are buggy.

Anyway, one thing I dont understand about bfast, is I keep getting this error:
In function "BfastIndexValidateInputs": Fatal Error[OutOfRange]. Variable/Value: indexNumber.
Message: Command line argument.
***** Exiting due to errors *****

I'm using the same settings as I used before for shorter reads without any problems? what should i change?

Indels are an extremely important source of variation, especially in cancer. Anyhow false indels do occur, but the read usually is still mapped to the approximate correct location. I developed SRMA (http://srma.sf.net) to clean up such indel errors.

Could you post your command you use with BFAST and I can take a look?

[Kevin_Johnson]

The problem is a typo in your book,

Perform local alignment:
$bfast-0.6.4d/bfast localalign -f hg18.fa -m bfast.matches.file.hg18.<N>.bmf -A
> bfast.aligned.file.hg18.<N>.baf


You just show -A but not -A 1

Also, is there a way that you could preload masks or have it so users can just choose a mask parameter rather than have to enter 10 different commands for one part of the workflow?

[nilshomer]

The problem is a typo in your book,

Perform local alignment:
$bfast-0.6.4d/bfast localalign -f hg18.fa -m bfast.matches.file.hg18.<N>.bmf -A
> bfast.aligned.file.hg18.<N>.baf


You just show -A but not -A 1

Also, is there a way that you could preload masks or have it so users can just choose a mask parameter rather than have to enter 10 different commands for one part of the workflow?

The typo has been fixed. Thank-you!

Since the indexes need only be built once, and you will have many experiments mapping to the same reference (hopefully), I don't think pre-loading them is necessary. I don't want it to be too easy for you

[bpetersen]

Does anyone know if PerM will support the new SOLiD paired-end sequencing data anytime soon?
In the manual it only mentions mate-pairs and I think it needs some adjusting to also handle paired-end, or am I wrong?
Thanks for your answers!