454 Sequence reads with the wrong MID??

[melcar]

I am analysing a environmental barcoding 454 dataset. As this is the first run I have completed I individually identified all species in all samples prior to bulk PCR and 454 sequencing, so I know what should be in each sample. While it has worked well I have one quirky thing with the dataset where I get 1-4 sequence reads appearing in samples that represent species that should not be in the sample. I though it may represent gut content of my species (as guts where included in the 454 extractions), but these reads are sporadically appear and are not alway associated with all replicates of a sample. I used a universal tailed design (with 2 PCR step to attached the MID) so I not sure how it could happen but is it possible for a small number of reads to be associated with the wrong MID? Has anyone else noticed this? Any ideas??

[JackieBadger]

I saw a paper in press that describes this source of error, but I forget the title and journal. it describes something in the way of MID swapping or errors in the MIDs that create either new MID combination not employed in the study or mimic existing MID combos. I have seen the same thing in our dual MID data i.e. we get a bunch of duel mid combination that were not employed, at very low sequencing depths.

Just ignore them in my opinion

[Anthony.287]

JackieBadger,
Any chance you could help me find that paper? We're dealing with a similar issue, and trying to rule out anything and everything is rather time consuming.
Thanks!!

[JackieBadger]

SURE..found it!

http://www.sciencedirect.com/science/article/pii/S1754504812000918

[JackieBadger]

One other thing that could explain your observations is aerosol contamination of PCR products, or during DNA extraction.
DNA contamination may occur if you are using a robot for extractions where the same pipets are only rinsed. For all next-gen library preps you should follow strict pcr-PCR conditions. i.e. separate rooms, barrier filter tips etc

[CG&R]

We consistently see something similar (8+ 454 Jr. runs).
Multiplex sequencing of samples that each have unique MIDs on 5' and 3' resulting in reads that have crossing of MIDs.

(samples prepared on separate days, using pre- and post-PCR rooms)

Occurring at about 1% of total reads.

Most likely this occurs during emulsion PCR by cross-priming of unincorporated primers (no matter how good your sample prep is).

Hope this is informative

[yaximik]

Just out of curiosity I used sfftools to split one of earlier sff files from multiplexed run. In addition to expected barcodes 1 to 5 I found reads with unused barcodes 6-12 at the rates of few percent each. Interestingly, tubes with barcodes 6-12 were never opened yet at the time the libraries were prepared and I am the only one in the entire institute who uses 454 sequencing. So the possibility of any sort of contamination is out of question. Either reagents were cross-contaminated already or there got to be another reason for emergence of reads with unexpected barcodes.

[JackieBadger]

Just out of curiosity I used sfftools to split one of earlier sff files from multiplexed run. In addition to expected barcodes 1 to 5 I found reads with unused barcodes 6-12 at the rates of few percent each. Interestingly, tubes with barcodes 6-12 were never opened yet at the time the libraries were prepared and I am the only one in the entire institute who uses 454 sequencing. So the possibility of any sort of contamination is out of question. Either reagents were cross-contaminated already or there got to be another reason for emergence of reads with unexpected barcodes.

Same goes here.
The paper I linked too earlier discusses.

[Anthony.287]

Thank you for the link! Just starting to get the hang of sfftools, and it looks like I'm finding the same...