HGAP assembly

[coldturkey]

I recently installed SMRT portal on a virtual server. We used Ubuntu 12.04 server as we have SMRT portal all all associated programs (SMRT view, HGAP etc) running well on a desktop machine which is also running Ubuntu 12.04.

I have completed a number of HGAP assemblies on the desktop machine and these ran without any problems. However, when the same jobs are run on the server everything proceed as normal untill the assembly stage. It appears that the reads are not being assembly or that step in the HGAP pipeline is not running.

For example the assembly.qc file (located in the assembly dir) only contains the sections

[Unitig Consensus]
[Contig Consensus]
[Read Depth Histogram]

no sections detailing information of Scaffold numbers, contig numbers etc are produced.

Additionally, HGAP assembly of 2 smrt cells (on the desktop machine) produces an assembly containing 200 contigs while on the server the consensus.fasta.gz file contains over 3000 fasta entries.

I am assuming that the the assembler is not running or at least not running correctly and the output is merely the corrected long reads produced in the first step of the HGAP assembly. Is this correct?

Have you encountered this issue before?

Regards

Brian

[krobison]

I'm having similar fun getting HGAP & Quiver to run correctly under Ubuntu.

In the directory you launched from, there is an sge_log directory. If any of these have serious error messages (can't find an executable or shared library) that may help

Good luck!

[rhall]

A consensus.fasta.gz file would not be produced unless the Celera Assembler stage had completed and the raw reads mapped back to the draft contigs. HGAP is deterministic so you should get the same results on the two machines.
The most common system issues causing the celera assembler step to fail are sge queue setup problems.
Can you post the ./log/smrtpipe.log for the job?