CLC assembly surprise

[nagual]

Hi everyone
Normaly we had used CLC, among others package, to make assembly of our 454 reads. However we had detected some rare result the last, we have a typical 454 run with reads filtered and with minumun length of reads of 50bp, however the output of CLC include contigs of 30bp. We have checked two times the parameters of the program and all is OK (including lenght minumun of the contig >50).
Anyone have some idea?

[prianda]

Dear nagual,

Feels like we're having the same issues. I'm dealing with Illumina RNA-seq paired-end data (2x72 bp). The shortest insert fragments in our library are app 100 bp long (as seen in a Bioanalyzer run), but the CLC reports display paired-end distances of 50 bp.

Moreover, the vast majority of my paired reads doesn't map to the same contig, even if the contig size is larger than the mean paired-end distance...

Have you had any luck solving this?

[GenoMax]

Have both of you contacted CLC tech support to get help. The problem description appears to indicate that this may be a software bug.

[prianda]

Dear GenoMax,

I´ve contacted the CLC support and explained the matter with the fragment sizes. But we could not reach any plausible solution...

[sklages]

Dear GenoMax,

I´ve contacted the CLC support and explained the matter with the fragment sizes. But we could not reach any plausible solution...

What is their explanation?

[prianda]

Dear sklages,

In their last message, they were wondering about repeated regions. They would be collapsed in the assembly and during mapping we would see a distortion in the positions of each of the paired reads.

[mchapman47]

Hi Nagual, CLC does a de novo assembly and then maps reads back to it. Is it possible that the de novo cut-off of 50 is being adhered to by the program but when you map the reads back theres poor coverage/quality on the ends and its trimming the final length to <50? We found a similar thing assembling a transcriptome and this is the only thing we could think was going on!