Entering edit mode
5.8 years ago
Inquisitive8995
▴
280
Hello,
I have 6 samples for demultiplexing. I am using bcltofastq. When I created a sample sheet for these samples and demultiplexed them, one of the samples (say sample1) had very low number of reads (About 5k). But when I removed all other samples from the sample sheet and then redid the demultiplexing with just sample1 , it increased the number of reads to 200k. I am surprised in how this happened and I cannot understand the reason behind this. Can someone please help me in understanding this ? Thanks in advance.
That shouldn't be possible. Try that again with the exact same commands each time and ensuring that only the sample sheet is changed.
I am changing only the sample sheet and the output folder. Everything else is exactly the same
Do you allow barcode mismatches?
The default barcode mismatch of "1" is allowed
There should be a folder
Statsand within a fileDemuxSummaryF1L1.txt. There you get a summary of found index sequences and there count. Compare the results. Maybe you see something siginificant.Is the result different if you don't allow mismatches?
fin swimmer
I will check those and try without mismatches also. Thank you so much
Only way this is possible is that once you remove other 5 samples some of the indexes (that are closer to those 5) are now getting assigned to this sample because of 1 bp mismatch allowed.
If you are concerned about mis-assignment then use perfect matches on index sequences.
bcl2fastq is usually picky enough that it will throw an error if any barcodes are within 2*mismatch though.
In a test I did having just one sample did not change the count of reads for it (8 bp index so more specific) whether treated singly or as a part of the pool.
Inquisitive8995 : Can you use the code in this answer to count the actual index sequences that show up in your two sets (done with other samples and without other samples)?
Thank you for the reply. I will try this code and get back to you .