So many software for alignment!!!

[found]

Hi,

I am new to the area of mRNA-Seq data analysis, actually also not "old" to the bioinformatics.
Recently, I have read a paper involved in the mapping of mRNA-Seq data to the human genome by ELAND. I am very interested in the mapping result and want to do some data mining for my own purpose. And the author only gives out the original mRNA-Seq data.
I have browsed this forum for days and learned a lot. first thing is that ELAND is not available for me. I found that someone have listed so many software for short read sequence alignment and I feel really confused about that which one is suitable for me. That's why I am here to get some suggestions from experienced expert.

My object is to map all of these mRNA-Seq to a database which includes the human genome and my own set of sequences(56bp). At most two mismatches and no gaps are allowed. Since a small part of sequences my contain symbol '.' such as "ATCTAT.CGTACG.GCTAGTGGTGAAGG", the software should not filter '.' out and consider it as a mismatch during alignment.

Can you recommend a software to implement this purpose. Of course, both accuracy and speed are important. I think accuracy is more important if they can not be satisfied simultaneously.

[doxologist]

IMHO, I think method with Burrows-Wheeler transform are now the most optimal.

Bowtie is quick and easy...
bwa in the MAQ package is more flexible and extensive.

Hope this helps.

[doxologist]

oh.. by the way... the prev recommendations is for data with Illumina and 454. If you want to analyze Solid data, different software must be used for colorspace. Visit the solid thread for more info there.

[found]

Thanks l lot. it is Illumina data.
when you mentioned "Burrows-Wheeler transform", is it a tool, or an algorithm? or it is just the full name of "bwa in the MAQ package"?

[jkbonfield]

Burrows-Wheeler transform is an algorithm, originally used for data compression in tools like bzip. The interest for bioinformatics is that with more recent tweaks (something called "FM Index") it can be both a compression tool and also an indexing tool.

This indexing tool is what makes it good for short-read aligners, bwa and bowtie being two such examples. Whether this method is optimal for you depends on the size of the reference genome you're comparing against.

As for why there are so many tools, well it's harder than you'd think. Sure it's easy to simply align data, but to do it for read-pair data with short indels taking into account multiple matches and computing probabilities that the sequence has been misaligned (etc) all adds up to a complex task. This has lead to a lot of competition between groups. I expect in time the number will dwindle as we get winners and losers.

[doxologist]

I agree... there would eventually be a few dominant tools for each type of information space... which would take the best algorithms and be the most data-friendly. The initiatives for common format would make this process much more efficient.

I think the future added value would no longer just be alignment, but what's downstream. SNP detection, paired ends, indel detection, handling large numbers of samples, etc.