Sample Cross-Contamination

[MAdkisson]

Hi All

I've been killing myself trying to find info on dealing with sample cross-contamination in RNA-Seq - lots of warnings to take precautions as cross-contamination is a real issue, but nobody talking about having experienced the issue and what was done to correct for it. Here's what I'm doing and what we found:

• Using TruSeq RNA prep kit with the 12 indexes that come with it
• multiplexing no more than 5 samples to a lane
• take the necessary precautions when ligating the indices to the cDNA samples
• doing full transcriptome analysis comparing single gene knock-out samples to WT, so a contamination issue of 1 in 1000 can have a significant effect. We have found quite a few samples with this level of contamination.
• several different tissues involved, and the contamination we're concerned with at this point is cross-tissue
• have a list of 5 highly expressed, ~very specific genes for each tissue against which we compare each sample to check for contamination
• we're almost certain that the cDNA prep step is the culprit as all the contaminating tissues were processed no more than 4 tubes away from the contaminated sample, on an 8 strip tube

My main questions are:

1. in what environment are others prepping their cDNA for sequencing and are any major precautions being taken to eliminate cross-contamination?
2. once data has been generated in which cross-contamination has been observed, are there any data manipulation fixes that anyone knows about or is this data simply garbage?

Thanks
Michael

[pmiguel]

If your experiment can be ruined by 1 in 1000 cross contamination, I think you are hosed. But here are some issues that spring to mind:

(1) Obviously never use anything but filtered pippetter tips. I presume you would already be doing this. That is, never stick anything but a fresh, filtered pippetter tip into a sample.
(2) Again obvious, but make sure to require zero mismatches to your index sequence during demultiplexing. You could consider moving to longer indexes if you suspect that the contamination is a sequence-error issue. You could also tighten up the quality requirements for your barcode read -- require quality values above a certain threshold for every base, or toss the whole thing.
(3) You probably want to invest in a good HEPA filter unit. Then test how well it is working by doing PCR before and after use on air samples. The issue here is dust carrying library molecules.
(4) You would need to be sensitive to capping/decapping issues that might produce aerosols.

Good luck.

--
Phillip