De novo assembly of a large insertion with a mapped-read seed.

Question

[Pierre Lindenbaum]

Short story:

We are trying to identify the content of a large insertion localized at a known place in the human genome. Do you know an existing software that is able to start and extend a de-novo assembly from a read mapped to the REF sequence?

Long story:

• My collaborators have sequenced the region of interest using Ion-torrent + mate-pair. As far as I can see, there are a lot of mismatches in the aligned reads.
• My BWA alignment returns a poor depth in the region of interest (~ 2 )
• CGH shows that the insertion is a junction of a head-to-head duplication

Answer 1

[Pierre Lindenbaum]

posting @BaCh_mira 's answer on twitter here:

@yokofakun Yes, MIRA. Used iteratively in mapping mode. Automated in MITObim: http://t.co/nCcVK3glwc https://t.co/xE20cUbKG5

— Bastien Chevreux (@DrMicrobit) October 17, 2014

Answer 2

[Devon Ryan]

Have you tried making a faux set of reads from the neighboring reference sequence and then seeing what trinity outputs when you try to assemble those along with the other reads? You could just blast the neighboring sequence against the resulting contigs to quickly narrow down which ones are candidates (assuming the results are worthwhile). Something like that might be worth a try at least.

[Pierre Lindenbaum]

moving to an answer for the discussion. I tried to run vevlet with the reads mapped in the ROI and the unmapped reads: no contig of interest was created (= large number of short contigs)