Suppose if I mislabeled some samples with wrong barcodes, for example : SampleA supposed to be designated with BarcodeA but ligated mistakenly with some other barcode, Is there a way to fix this problem bioinformatically.
I found several of my samples mislabeled this way after i split the pool using barcode splitter, when i did not find any reads corresponding to those samples and i am pretty sure that i have added those samples in the pool.
Is there a way to know what barcode went into what?
We QC our data by checking what barcodes are present versus what was expected. We have also been using a second technology to do simple genotyping assays on a few SNPs for all samples entering the lab to allow us to check sample identities against the NGS results. Especially for clinical samples, it is very important (obviously) to minimize sample confusion. I agree with Josh Herr, though, that salvaging mishandled samples is not always possible or even desirable.
I'm not sure what exactly happened here and if you can fix the situation, you didn't give us a very clear explanation of where exactly the error took place: in the wet lab or bioinformatically. In order not to contribute to bad science: when in doubt, throw it out!
Josh, I'm sure the OP thought of this already and is really asking whether there might be a way of informatically handling a common mix-up (who hasn't swapped samples & barcodes? It happens...) without tossing what is probably a several thousand $ experiment. Asking a question doesn't imply the OP is "contributing to bad science."
I do apologize if my comment came across as snarky, that was not my honest intent. That was before the question was clarified and proceeding the "i don't know which samples are which" question, so I was just commenting on I how I think it's a bad idea to assume. I apologize if what I said implied that anyone was "contributing to bad science" and barcode mixups have certainly happened to me before, so I understand how this can happen to anyone.
This is a fairly common error that we see. In such cases you can split by all possible barcodes supported by the platform (a few dozen usually). The barcodes are designed to be maximally different from one another plus some other considerations. The sequencing instrument has documentation that lists them all. Usually you won't even need to process all your data just a smaller subset of it.
Of course the usefulness of this may be limited. But if only one or two barcodes are missing you can recover them this way.
If you have custom barcodes then you can just cut out the barcode region from each read and count the number of times they occur. This will also be a good indicator of what barcodes are present.
Sorry if i wasn't clear. The problem occurred in the wet lab.
Josh, I'm sure the OP thought of this already and is really asking whether there might be a way of informatically handling a common mix-up (who hasn't swapped samples & barcodes? It happens...) without tossing what is probably a several thousand $ experiment. Asking a question doesn't imply the OP is "contributing to bad science."
I do apologize if my comment came across as snarky, that was not my honest intent. That was before the question was clarified and proceeding the "i don't know which samples are which" question, so I was just commenting on I how I think it's a bad idea to assume. I apologize if what I said implied that anyone was "contributing to bad science" and barcode mixups have certainly happened to me before, so I understand how this can happen to anyone.