[Nomijill]

This is great thank you.

[joa_ds]

Hi there, i am under some pressure to deliver results from a solexa run. They did a run, and all i had to do was give them a list of variants.

Problem is, i ran bowtie, give them the list. They verify with Sanger sequencing, and the results are different. So i used MAQ then. Results different than BowTie results (but also different than the 'reality' obtained with Sanger).

So I have a lot of questionmarks... If results vary depending on the mapping software, well, my neighboor's genome has not changed since i pressed the MAQ button, so that is a problem.

You guys still think using Eland will give me some 'strength' in explaining i did the best possible mapping i could do?

Problem is that they have multiple genotypes mixed and are looking for QTL's, so a 40% or 60% variant, in the 'normal world' is just a heterozygous variant, in QTL world, it is a big difference...

[der_eiskern]

Hi lh3,

Thanks for the original benchmarking! I'm actually looking for such a benchmarking including the latest tools, like ZOOM!, bowtie, R Biostings pairwiseAlignment(), etc. Has anybody heard of that?

Cheers,

N.

I second this. I'd be interested in a comparison of erroneous mappings with the dominant WGS alignment-to-ref tools.

[dan]

I'm definitely out of my league in this discussion, but if anyone needs hosting for some of these sample datasets, let me know!

Can we make a wiki to store all the datasets and results? A nice table would be soo cool! Also, I think people contributing data and analysis to a wiki page is more 'cite-able' than forum posts, although this is clearly an effective mechanism for community discussion. A wiki page to summarize this thread could be a good 'citation'. I tried to find a suitable spot on Wikipedia, but they don't even have an article for next-generation sequencing yet, never mind short read aligners to cope with the data.

I'd like to start benchmarking the de-novo aligners (where accuracy is much harder to pin down). This will become more and more of an issue as all the genome sequencing projects move over to next-gen technology.


BTW, anyone seen this:

I’ll be attending the coveted Marco Island meeting early next month (February 4-8), where I’ll present a poster on my evaluations of short read aligners for next-gen sequencing data

http://www.massgenomics.org/2009/01/short-read-aligners-update-at-agbt.html

Looks relevant to the topic of this thread.

[regyre]

Where to find a recent benchmarking?

Hi lh3,

Thanks for the original benchmarking! I'm actually looking for such a benchmarking including the latest tools, like ZOOM!, bowtie, R Biostings pairwiseAlignment(), etc. Has anybody heard of that?

Cheers,

N.

[valeu]

Hi Heng!

Have you heard about SOCS (http://bioinformatics.oxfordjournals...tract/btn512v1, http://socs.biology.gatech.edu/)? In their article they say that "The overall algorithm is similar to that used by software tools developed for analysis of Illumina-Solexa data (Li et al, 2008; Smith et al, 2008)

I'm interested in alignment of SOLiD data and I'd like to know your opinion what to use.. Maq, Mosaik, SHRiMP, ZOOM or this new tool SOCS ..

Best regards,
Valentina

[jhui]

SeqMap (http://biogibbs.stanford.edu/~jiangh/SeqMap/) - work like ELand, can do 3 or more bp mismatches and also insdel

[BaCh]

...
MIRA
...
Although the version 2.9.95 doesnt support SNP analysis yet but its compact.
...

It does since 2.6 or something. Alas, the docs need to catch up.

[zee]

I am keen to find out what the optimum set of parameters are for MAQ in situations where we expect to find more indels as welll as the 0-2 mismatch hits.
The latest version sounds like it would be better for minimizing false positives.

[ShaunMahony]

Heng,
If you wanted to test RMAPQ, you could always convert the FASTQ files into an approximation of the PRB files:

http://seqanswers.com/forums/showthread.php?t=236

Shaun

[cgb]

Eland is going to be hard to beat. It has had a few years of hard work, optimisation and thought put into it by Anthony (SSAHA) Cox at Solexa. It was designed from day 1 for aligning far more reads than you currently get from a GAII, to a full human reference on a desktop computer.

[lh3]

-H FILE dump multiple/all 01-mismatch hits to FILE [null]

If you specify this option, maq will dump all hits to a gzip file. -C specifies how many hits to output.

[myrna]

The latest version, 0.6.6, can output ALL hits with 0- or 1-mismatch in the seed.

Hi Heng.
I was excited to see this feature added to MAQ in the latest version as much of my work is applied to RNA (hence, it is quantitative). This should (hopefully) allow me to reduce some biases introduced by losing reads which map to multiple locations. Now, I am wondering, how do I go about using this feature? Is there a new option when running maq map? Or mapview? I have been unable to find it in the manpage.

Thanks,

Ryan

FOLLOWUP:

I found out the answer to this in the latest doc provided with version 0.6.6 (not the version on the sourceforge page).

Usage: maq map [options] <out.map> <chr.bfa> <reads_1.bfq> [reads_2.bfq]

Options: -1 INT length of the first read (<64) [0]
-2 INT length of the second read (<64) [0]
-m FLOAT rate of difference between reads and references [0.001]
-e INT maximum allowed sum of qualities of mismatches [70]
-d FILE adapter sequence file [null]
-a INT max distance between two paired reads [250]
-n INT number of mismatches in the first 24bp [2]
-M c|g methylation alignment mode [null]
-u FILE dump unmapped and poorly aligned reads to FILE [null]
-H FILE dump multiple/all 01-mismatch hits to FILE [null]
-C INT max number of hits to output. >512 for all 01 hits. [250]
-s INT seed for random number generator [random]
-N record mismatch positions (max read length<=55)
-t trim all reads (usually not recommended)
-c match in the colorspace

[Amit]

Mira

Hello everybody,

I am new to this group and couldnt resist myself to join this exiciting discussion

Has anybody heard of MIRA ?

http://www.chevreux.org/projects_mira.html

There is this guy quitely working on another software tool for Next Gen assembly .

The USP of this tool is it can perform a true hybrid assembly SAnger+454 or 454+Solexa which I believe will solve the Next gen assembly issues.

Although the version 2.9.95 doesnt support SNP analysis yet but its compact.

This tool might be on slower side becuase it performs assembly iterative correcting errors on the way.

I hope somebody evaluates this new version becuase I dont have the much needed hardware to run this program.

regds,
Amit

[lh3]

The latest version, 0.6.6, can output ALL hits with 0- or 1-mismatch in the seed.