next_phred & Phaster - call for saving images from GA2

[bioinfosm]

This was a very interesting read, slated for release in 2010, Phil Green presented some benchmark data

http://www.bio-itworld.com/news/10/05/09/Phred-software-data-Illumina.html

Can anyone share the actual presentation?

With images compressed 75% using an open tool, it will be interesting to go back to the debate of how much to invest in data management. Also, other base callers like pyro-bayes and SWIFT are out there, not sure how they compare, and what happens with Illumina's major release of semi-ordered arrays early 2010 (thats what I heard)

sm

[Torst]

With images compressed 75% using an open tool, it will be interesting to go back to the debate of how much to invest in data management. Also, other base callers like pyro-bayes and SWIFT are out there, not sure how they compare

My experiments in 2008 showed that you could compress the images losslessly using the open JPEG2K-LS software to about 50% of their original size. This was Illumina GA1 data at the time. I have not re-tested since upgrading to GAIIx.

PyroBayes is for base-calling of .sff flowgrams from 454 GS machines. It is not for Illumina data.

That article was not written very well. Some of the numbers/percentages quoted are unclear what they relate to. No mention of de novo assembling (a la "phrap"). It is not hard to map reads better than Bowtie. Other mapping tools like MAQ do use quality information.

As for extracting more reads from the TIFF image, at a per pixel level, I suspect there is some information to be gained, but maybe not the 250% claimed. At the moment the GA Pipeline is effectively calling a read per bunch/cluster of pixels. Sure, you could call individual reads from pixels within the clusters (if well aligned) but they will have lower quality. So the current "read" can be considered a "consensus" read of those produced by next_phred.

I look forward to anything Phil Green produces, so let us hope more information becomes available soon.

[john_mu]

any update on this?

[Torst]

any update on this?

I have not heard anything since. I suspect the improvements made in the GA Pipeline 1.4, 1.5 and 1.6 made it not worth developing further - but that is just a hypothesis.

[bioinfosm]

I will try searching, but like Torst, seems not much of resources being put to developing that part