Demultiplex dual index NextSeq run with only one barcode

Question

[blackjack666]

Hi everybody!

I am not very experienced with sequencing, so maybe there might be a simple solution to the following problem. We ran a sequencing run on an Illumina NextSeq this week with 75 cycles and dual indices. The original plan was to load 12 different library samples, each barcoded individually with a single 8 nt barcode and then demultiplex using bcl2fastq. This approach has worked before. This time, during setup of the run we chose dual indexed sequencing by accident and noticed when we tried demultiplexing using bcl2fastq. At the moment I demultiplexed without using a sample-sheet, so I have a large fastq file containing undetermined reads. Is there any way to demultiplex a dual indexed sequencing run when only a single barcode per sample was used? Any ideas using a script would also be very welcome, programming is no problem for me.

Answer 1

[Devon Ryan]

You can still use bcl2fastq, you need to just mask the second barcode (e.g., --use-bases-mask Y*,I8,nnnnnnnn,Y*). We do this all the time for our HiSeq runs, since we have different barcode strategies on different lanes.

Answer 2

[Gabriel R.]

Essentially, you have only the information provided by a single index instead of two. We published a maximum-likelihood demultiplexing algorithm that can incorporate uncertainty due to missing cycles, missing indices, poorly designed indices and high error rates:

http://grenaud.github.io/deML/

paper:

deML: robust demultiplexing of Illumina sequences using a likelihood-based approach. Gabriel Renaud, Udo Stenzel, Tomislav Maricic, Victor Wiebe, Janet Kelso Bioinformatics. 2015 March 1.

Answer 3

[Benn]

The fastx toolkit has a barcode splitter, see here