Picard demultiplexing with random barcodes

Question

[RNG_Daemon]

I am demultiplexing RNA-Seq data using Picard. I normally fill out the library_file with BARCODE_1 for the sample barcodes and BARCODE_2 for the UMI barcodes.

However, in this run, I don't know the UMI barcodes. They came from a pool of random barcodes and are 10nt in length.

Now I don't know how to fill out the BARCODE_2 row in my library files. Leaving it empty, filling it with "N" or "NNNNNNNNNN" lead to errors like Row for barcode null appears more than once in LIBRARY_PARAMS or BARCODE_PARAMS file.

Has anyone worked with this?

[i.sudbery]

Can I ask why you wish to demultiplex using UMIs? Normally one would use UMIs together with an UMI aware deduplicator, such as UMI-tools, to remove PCR duplicates, rather than demultiplexing into different files.

[RNG_Daemon]

Using umi-tools was indeed the right thing to do. Thank you!