Newbler 2.3 on Ubuntu- Anyone?

[WaltL]

I'm running Newbler 2.3 on an Ubuntu box (MacPro, dual quad core, 32 Gb RAM) and having some issues. Wondering if anyone out there is using Ubuntu and experiencing similar problems?

Everything appears to be working w.r.t. the GUI and the app., especially if I run in genome assembly mode as a complex genome, which seems to work fine. However, when I try to run in cDNA mode (I'm doing de novo transcriptome assembly) it takes forever on even relatively small jobs (2-3 gig 454 data). The exact same data run on a RedHat install takes ~30 min.

If I try to do a hybrid assembly with ca. 5xe6 seqs (12 gB data), it always stalls during detangling, ~ 55% complete), loses contact with the GUI, and locks one of my cores @ 100% and my RAM @ ~ 14 Gb. I can exit the app., but I have to reboot to free up the core and the RAM.

Any thoughts/suggestions as to why the cDNA algorithm won't run properly on an Ubuntu install? Roche tech support is working with me, but so far no fixes.

[yvan.wenger]

Hello,

I*am experiencing the exact same problem (cDNA, also MacPro, dual quad core, 32 Gb RAM on ubuntu). I did not have the problem with newbler 2.3.

The only difference is that the behaviour similar in Red Hat.

Did you find any reason?

Best,

Yvan

[WaltL]

Yvan,

Wish I had some good info for you. I'm still running Ver 2.3. My fix was to filter the data outside of Newbler and then do hybrid assemblies using fasta/qual data. Prior to that, I was loading .sff files for the 454 data and fasta/qual files for the Sanger data plus loading my sequencing filtering file for ribosomal, mito, etc...

Using sff_extract on the 454 data and then filtering all data using a SeqClean pipeline prior to running Newbler (and also without running a filtering file) sped things up significantly and prevented the problems I was experiencing.

So, I'm not really sure which change helped the most: filtering prior to assembly or assembling only fasta/qual data.

I'd be interested to know if you discovery anything further.

Walt