Errors in barcodes

[Anthony.287]

Hello, again!
I’ve been getting into the informatics side of 454 sequencing more lately, and have some questions. Mostly about barcodes…
1. Does anybody look for barcodes that shouldn’t be in a dataset? If so, how often do you find them?
2. Everything I’ve read lists an error rate of ~1% for the 454. Does this sound right?
3. Is there a normal error rate for barcodes?
Basically, I have one set of samples, using older, shorter barcodes, and keep finding other barcodes within this set. They occur at a low rate, ~.2%, but I have no way of knowing if they are real sequences or if they are occurring because of PCR errors.
Any help or advice would be greatly appreciated! Thank you!!

[jswebb2]

Bar code problems

Is this with extended read runs?

[Anthony.287]

If you mean the FLX+, then no. We're using the FLX titanium, with software 2.5.3.

[JackieBadger]

It's definitely a real phenomenon in 454 data.
If you are seeing totally new barcodes then it could be systematic base mismatch error. This form of error is sequence specific. I have seen it a lot in 454 and Illumina data sets where base mismatch errors continually occur at particular positions. These erros can occur at a higher rate than ~1%. For example in samples that I may have sequenced at around ~4000x I would be very surprised if the total sum of numerous systematic base mismatches did not add up at at least ~200x. I have seen systematic base mismatch errors occur in the 100sX and systematic indel errors occur at higher depth than its parent sequence.

If you are talking about newly formed combinations of MIDs then thats a different but very real issue. It is not a problem when you see combinations of MIDs that you did not employ, but the same source of error likely causes cross contamination between samples if you will.
We spoke about it previously: http://seqanswers.com/forums/showthread.php?t=24668