Illumina1.8 Paired-End Barcode Splitting?

[pbatzel]

Hi everyone,

Hopefully this is a pretty simple question with a pretty simple answer. I've been using a pipeline for pre-processing Illumina1.8, fresh off the sequencer, data. The pipeline puts sequences with the correct first 5bp (we have a list of 5bp barcodes denoting different samples), including however many mismatches we want to allow, into separate files. We run it first on the /1 half of the data and then on the /2 half and end up with AACCC_1.fq and AACCC_2.fq files for each different barcode.

My question comes in when a given sequence from the /1 side matches a barcode in my list exactly, but the /2 side is not the same as the /1 and doesn't match any barcode. Am I able to safely assume that because the /1 and /2 have the same sequenceID from Illumina and are supposedly read from different ends of the same molecule, that I can put the /2 into the same barcode file as the /1?

My biology background isn't very strong, so sorry ahead of time is this is a simple question. Right now if only one side matches a given barcode, the other side is just thrown out. That would be a terrible waste of PE data if my assumption above is safe. Let me know what you think.

[swbarnes2]

I guess you are running your own custom barcodes somehow? Like they are embedded in the sequence at both ends?

But in theory, if the reads came from the same flowcell, and lane, and tile, and xy-coordinates, (which is all in the original name given by Illumina to the read)they should be the two ends of the same DNA fragment.

[pbatzel]

Hi swbarnes thanks for your reply.

Yes, not knowing the biology behind it, the barcodes are somewhat magical to me (although I suppose a lot of this is). My very rough understanding of it is that we use a specific primer for each different sample and therefore can set our own 3' and 5' "barcodes" or adapters, or whatever else someone might call them. I also should have mentioned this is RNA-Seq stuff, so everything is cDNA.

You're spot on with the illumina "ID" I mentioned being something like lane, title, xy and I am indeed hoping that I can make the assumption that they are from the same DNA fragment if they have the same ID. If they are from the same fragment and one side has a barcode that is correct, I should be able to fix the barcode on the second side and bin it into the correct file.

After I posted this question, a colleague mentioned something about "chimeric molecules". I'm assuming from the PCR step, that would botch up this assumption, but only for a very small % of the reads unless something has gone horribly wrong. Again, I'm very new to all of this and have very little biological background so any info is very helpful.