I am assembling Illumina sequences (MySEQ paired end 250+250 bases) obtained for bacterial genomes. I used both Velvet and Abyss software but I did not obtained a quality files for the genome assembly (a file with the quality of each base in the assembled genomes).
I have previously submitted genome sequences to ncbi (obtained with 454 FLX sequencing) but the quality for the final assembled sequence was compulsory required for the submission. Does anyone have any suggestions? Is it possible to obtain quality files for these Velvet/Abyss assemblies?
I don't know about Abyss, but Velvet does not return quality files for the assembly. .I belive that you can submit the genomes without quality files, because every genome nowadays is done using new sequencing technologies (like Illumina).
You can check the quality of your assembly aligning the original reads against the contigs (you can see indels or assembly erros), but you can't measure that in a value.
If you really need the quality values, you can use other assemblers in your data. I have used the gsAssembler (from Roche) in Illumina data with success.
Regards
Velvet actually ignores the quality values for reads, which is why I normally just use fasta read files that are pre-trimmed for quality. However, I might not have completely up to date information so you should send this query to velvet users e-mail list (for details see http://www.ebi.ac.uk/~zerbino/velvet/). They are very responsive. On the subject of NCBI, are you sure you need quality information for assembled genomes? I did not need to do this last time I submitted something.
So, this isn't quite the answer you'll want:
Most assemblers don't return q values, either for the individual bases, or for the overall assemblies. We (LANL), have been working on a tool that assigns quality values to assembled contigs after the fact, based on read-mapping. It's based on several assumptions, though, including cutting off quality at Q40, rather than the more traditional Q60 for final assembly. If you're interested in what our tool does, feel free to message me directly.
As far as I understand, the quality of the sequences is information coming out from the sequencer data during the sequencing process, it has to do with experimental details.
Low quality reads are hardly any help so you must filter them out before assembly. The assembly problem is an NP-complete problem and the De Brujn Graphs give you an approximate reasonable solution, not the optimal solution. Thus, the more noise in your data, the higher the chance for assembly errors, information loss, etc.
If you want to figure out how your reads contributed to the assembly you should map back your reads to the final contigs (using bwa, bowtie, or other fast aligner), and then you can sort mapped, unmapped reads, and unused reads.
Best wishes
Rogelio
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics