Identifying (and removing) 454 homopolymers frameshifts

[lionelguy]

Hi!

I have resequenced two strains of the same bacteria using 454 data. It seems that the run was rather crappy, and I got a lot of homopolymer errors (in average 1 frameshift per gene). I need to get rid of those errors, without doing any further sequencing. My purpose is to compare substitution rates in these strains, so I don't really need extra-accurate sequences.

For the moment, my strategy is the following:
- Use the reference strain genes as a query with BLAST-Extend-Repraze (BER, http://ber.sourceforge.net/) tool.
- Parse the output of this (which seems not quite straightforward).
- Remove extra nucleotides in my sequence, or add Ns.

Shortly, BER takes the output of a regular blastp of your favorite protein, retrieves hits that match nicely against it, goes back to the DNA sequence corresponding to your protein (i.e, the CDS + some flanking sequence).

Is there any other tool that I'm not aware of? Do you know any BER output parser? Does my strategy make sense to you?

Thanks for your advice and comments!

[Mathgon]

I have the same problem with errors in homopolymers. Have you finally found a tool to correct them in comparison with a refrence genome?

Best regards

[SeaJane]

AmpliconNoise http://code.google.com/p/ampliconnoise/downloads/list is the best tool I've found for removing homopolymer error.

[mbakker]

running AmpliconNoise

SeaJane, have you been able to get AmpliconNoise up and running?

[colindaven]

I can install AmpliconNoise, yet all the steps for running it do seem daunting, especially with the number of datasets I have.

Has anyone got any scripts etc for automation available ? After all this is a fairly common task for 454 users.

[aartacho]

HMM-FRAME: 'state of the art' in pyrosequencing frameshift correction and protein domain classification (includes 454 error model in Viterbi HMMER algorithm). Unlike AmpliconNoise, it only corrects coding reads

http://www.biomedcentral.com/1471-2105/12/198