SamToFastq fails because of unpaired mates about blr

[FrickTobias]

In an optimal solution, duplicates would not be removed here because of the problem you correctly point out with single reads out of pairs.

However, currently read duplicates are used for finding barcode duplicates (aka where two barcode has tagged the same molecule). This leads to two barcodes sharing a lot of proximal read duplicates which subsequently are merged into one barcode for all reads sharing that barcode sequence.

The actual procedure doing this is first to remove all duplicates with the BARCODE_TAG=BC option and afterwards analyze duplicate reads left in the file with cluster_rmdup.py (by first marking duplicates without the BARCODE_TAG flag).

I suggest that I rewrite the script to so it ignores duplicate reads at the same positions if they do not have different barcodes, essentially compacting the duplicate removal (BARCODE_TAG=BC), duplicate marking & cluster_rmdup step into duplicate removal & cluster_rmdup.