Please suggest a good assembly pipeline for an ION torrent data for a bacteria. I have tried MIRA3 where the largest contig size is 1.1 mb, but N50 is some where around 7000 only. Tried to blast the obtained contigs with the 16s rRNA of the bacteria (obtained by capillary sequencing), but was not able to get any hit. Please do suggest something that can be used. Also tried BWA_SW to ap the data to its very close known genome, but out of 2.6 million only 7500 reads got mapped with a lot of gaps in between. Expected genome size should be somewhere around 2.5-3 mb (considering all closely known genomes).
You can get a two week free licence of a commercial software called CLCbio (http://www.clcbio.com/products/clc-genomics-workbench/). I very much liked the de novo assembly tool and perhaps it is also helpful to you!?
Hello,
We had good results with SOAP http://soap.genomics.org.cn/ and Abyss http://www.bcgsc.ca/platform/bioinfo/software/abyss/
MIRA works pretty well with Ion Torrent data and bacterial genomes. I'm using MIRA on a regular basis on different species and getting very good results.
First of all, which MIRA version were you using (latest dev builds >3.9 comes with a lot of improvements for PGM and MiSeq). Also a low N50 does not necessarily reflect a bad assembly. How many 'small' contigs were produced and were the coverage (per base and genome) sufficient? Also MIRA is not that intuitive to use so the problem could also be parameter related. Anyway, MIRA does have a very responsive mailing list where you can ask for specific help.
Judging assembly performance by trying to re-map 16S rRNA operons may also led to false conclusions as all assemblers (not only MIRA) do have problems handling long repeats. So it might just be that -also depending on several other aspect- the 16S operon region couldn't be assembled at all. Why not searching for all genes and see how many could be assembled.
Also that >99% of you data could not be mapped to a very close related reference indicate that there is either something wrong with the data or the reference is not as close as you think.
I don't think so. This is Ion torrent data. Moderately longer read these are.. so...velvet may not be the choice
mostly, read lengths are not equal... so Velvet's de brujin would be complex...
The graphs in velvet are based on k-mers not reads. It might be worth a try, what have you got to loose. Anyway I'm not sure you can really expect get a complete genome from such an assembly. You already have one contig that by your estimate is a third of the genome, so I would count that as pretty good. Most complete genomes or scaffolds reported are completed by PCR after an assembly is done. If I were you I would just annotate the draft and see if you have homologs for the majority of genes in related species - this would give you and idea whether you are missing anything that "should" be there. As others have said the 16s operon is often tricky to assemble.
- Badges
- Science topic
- Similar topics
- Biology