De novo amplicon reassembly?

[Elizabeth Bent]

Hi, we would like to use a MiSeq to analyze amplicons (for barcoding projects- so there are thousands of slightly different sequences for each amplicon) longer than 300 bases, using Nextera library preparation kits. The tagmentation step in the Nextera kits results in fragments shorter than the original amplicons that can then be sequenced.

My question is, has someone devised software appropriate for re-assembling these amplicons, when there is a mixture containing thousands of different original amplicons, while avoiding misassemblies? We also probably need to use de novo methods of assembly since we do not have every sequence in our database that might be present in our (unknown composition) samples, so we don't have an appropriate reference for every sample. Because these sequences represent the same gene region from different organisms, and there are conserved regions, we are concerned about chimeric misassemblies from the data.

Liz

[ians]

454 may be a better platform if you need more bases.

it will be difficult to de novo assemble all your amplicons given that they are very similar. For the ones you do know, you could align reads very strictly to each amplicon, and then assemble to get the full one back.

[JackieBadger]

Is your "amplicon" consisting of the primers, adapters, and indexes?
If your total amplicon is 300bp then you could do it.
You will have to wait towards the end of the year to get the upgrade from 150bp paired ends (total 300bp), to i think 250-300bp paired ends.

We have an amplicon, of which the total is under 300bp. We also know the exact length of our target region. Using pair-ends we can sequences from both ends, and stitch the mate pairs back together again using the over lapping region. I have listed known programs to do this below.
Stay away from de novo assembly in my opinion.

http://code.google.com/p/ea-utils/wiki/FastqJoin

Stitch https://github.com/audy/stitch

fastq-join FLASH http://www.cbcb.umd.edu/software/flash/

mergePairs.py http://code.google.com/p/standardize.../mergePairs.py

[JackieBadger]

One thing I didnt mention.
You mate pair sequences will be in two corresponding lists in separate FASTQ files. Each sequence will be ID'd by an index (sample ID), and coordinates for the cluster in which each ampicon is sequenced. In theory you could use these coordinates to put your mate pairs back together again, but I think most programs use the ordering of sequences within the corresponding sequence lists to join mate pairs.
So ones you have re-built amplicons, then you can sort my indexes.