Question

Why Pash 3.0 has been chosen for Roadmap Epigenome?

0

Entering edit mode

9.7 years ago

Saulius Lukauskas ▴ 540

I am going through the Roadmap Epigenome alignment protocols, and they claim they have mapped "a total of 150.21 billion sequencing reads onto hg19 assembly of the human genome using Pash 3.0 read mapper" .

Does anyone see the reasoning for using Pash 3.0 -- which is quite an odd choice given availability of more widely adopted aligners such as Bowtie or BWA? The latter two also support multithreading, something their aligner of choice lacks.

alignment pash roadmap-epigenomics • 2.6k views

ADD COMMENT • link updated 2.5 years ago by Ram 44k • written 9.7 years ago by Saulius Lukauskas ▴ 540

3

Entering edit mode

I don't know with any certainty (thus this is a comment), but I'd guess that Pash being able to handle normal and bisulfite treated read alignment played a role. Then at least there's just one single tool that can used for everything. I wouldn't have chosen that, but that's at least an argument in favor of Pash.

ADD REPLY • link updated 2.5 years ago by Ram 44k • written 9.7 years ago by Devon Ryan 104k

Ram · Answer 1 · 2015-03-18

2

Entering edit mode

9.7 years ago

Chris Miller 22k

The simple answer is that PASH is developed by the Milosavljevic lab 1, which is heavily involved in the consortium.

That was my PhD lab, and I helped a little bit on the PASH paper, so can report that it's a decent enough algorithm. I haven't seen any benchmarking against the next gen of BS aligners recently, so can't offer any up-to-date insight on how it compares.

ADD COMMENT • link updated 2.5 years ago by Ram 44k • written 9.7 years ago by Chris Miller 22k

0

Entering edit mode

Hi Chris. I'm seeking to use the roadmap RNA-Seq data to identify and analyze splice sites active in each cell type. Is there some way to get these from the PASH output? Perhaps something analogous to the BowTie/TopHat Junction files? Right now I only seem to have access to PASH .bed outputs and am not familiar with the interpretation of columns 4-6 (e.g.)

chr1    22336244        22336343        HAL:1220:C0NTJACXX:3:1112:2443:33889    1       +
chr1    22336249        22336348        HAL:1220:C0NTJACXX:3:1112:3363:36331    1       -

Thanks!

ADD REPLY • link updated 2.5 years ago by Ram 44k • written 8.5 years ago by bgulko • 0

0

Entering edit mode

Sorry, but I am not the right person to be asking about this at all. It's been 6 years since I used PASH. You're probably going to want to a) post this as a top-level question or b) email someone associated with the ongoing development of the algorithm.