indels using single end short reads!

[bioinfosm]

We had a sample with known 4bp deletion, but no tool would help me detect that...

any suggestions?

SSAHA supposedly does gapped alignment, but it gave me some 'novel' 1 or 2 base indels... not the one we know

[swbarnes2]

We had a sample with known 4bp deletion, but no tool would help me detect that...

any suggestions?

SSAHA supposedly does gapped alignment, but it gave me some 'novel' 1 or 2 base indels... not the one we know

SOAP may do it...it seems when you compile it, you specify how large a gap you are allowed to call for in the command line.

"3) Maximum gap size
-DMAXGAP=3
Maximum size of a gap allowed in a read, then "-g" option during running should not exceed this definition."

On the home page, they show 3 as an example, but 4 might work. I don't know how much it will slow down SOAP to allow it to try large gaps.

I know it finds plenty of 2 bp insertions when I use -g 2.

[BaCh]

Indels with 4 bases are on the border of what I would consider "sane" when aligning/assembling short sequences. E.g., a 36mer aligned against the same sequence but with 4 bases deletion gives you a score ratio (= score/expected_score) of barely above 70%.

I normally allow only 1 or 2 errors in Solexa mapping assemblies, but I quickly hacked together a change that will allow you to find indels or base changes with up to 4 bases in a Solexa mapping assembly. Grab http://www.chevreux.org/tmp/mira_2.9...x86_64.tar.bz2
and run the Solexa demo. Have a look at the results in gap4 and decide for yourself whether this would fit your needs.

Warning: Work in progress. Works for me, but not necessarily for you

Regards,
B.

[Torst]

myrialign

Maybe MyriAlign would be of use to you?

http://savannah.nongnu.org/projects/myrialign

[bioinfosm]

SOAP worked nicely on the data... Thanks to the person who shared his script to use soap results and generate indel calls

I was able to see the 4bp known deletion in the sample

Torst - are you the author of Myrialign? I will check it out as well

[mchaisso]

Depending on your coverage, you can try assembling the reads, then simply blasting the contigs against the genome. I know of a few groups trying to do this, but I haven't heard of success, so I'm curious if you try this how far you get.

-mark

[sparks]

Aligning with Indels

I've just finished a new aligner that will do indels up to 7bp. I don't have a web site for downloading it but if you'd like to try email novoalign @ gmail.com and I'll send you a copy. It's also at least as speedy as the best of the other aligners.

[ECO]

SOAP worked nicely on the data... Thanks to the person who shared his script to use soap results and generate indel calls

I was able to see the 4bp known deletion in the sample

Would said person be willing to share the scripts for using soap results? thanks in advance.

[sparks]

Novoalign and novopaired will do gapped alignments and is a fair bit faster than SOAP.
I've just released V1.03, this update improves quality scores for novopaired and also fixes a illegal instruction fault reported by one user.
You can download at www.novocraft.com
I've also changed the license term so it's free for any non-profit even if you don't publish in open journals.
Colin

[bioinfosm]

Would said person be willing to share the scripts for using soap results? thanks in advance.

Sorry but I never noticed your message in the new posts!

Sure, I would be happy to share. I used the soap algorithm, and then used a parsing perl script to get the results.

soap -a input -d reference -o prefix -s 10 -g 4

The parser is modified from Liu's script (BGI). You may PM me, and I will mail that to you, but would not want to put it up here..

sm

[kmay]

We had a sample with known 4bp deletion, but no tool would help me detect that...

any suggestions?

SSAHA supposedly does gapped alignment, but it gave me some 'novel' 1 or 2 base indels... not the one we know

Hi!

Glad to read that you managed the task. Is it from a mammalian genome? If so, would you be willing to share your data set with us ( of course NDA can be done)?
We would love to test our mapping on that challenge.

Klaus