Barcode Sequencing Errors

[Austin-s-h]

Hello again, I was wondering if you had a more elegant way of handling barcode sequencing errors. Currently, my snap object reports I have 85,228 unique barcodes. However, I know the number of true barcodes is 670.
Since I thought this was the result of sequencing errors in the barcode, I made a custom python script that adds the barcode to the front of the read header (because I'm starting from demultiplex'd fastq's) and any invalid barcodes (barcodes with >1bp mismatch) are renamed to INVALID.
When I perform simple filtering in SnapATAC (R), for a UMI count of 500 and a mit.ratio of <0.3 I lose almost all of these invalid barcodes (since there is a low chance of a sequencing error barcode having 500+ unique reads) and end up with ~1500 barcodes.
Is there a better way to handle these errors? Should I perform more stringent filtering? Here is what my current distribution looks like.

Thanks!

[Austin-s-h]

Actually, I followed through the tutorial just a little bit more and found that the UMI cutoff should be much higher, after that, it is 580 barcodes, which is exactly in the range I expected. Thanks!

[r3fang]

no problem. Another parameter for filtering barcodes is fragments-in-promoter ratio (0.2-0.8) which is also quite important, make sure check that when analyzing your data

…

-- Rongxin Fang Ph.D. Student, Ren Lab Ludwig Institute for Cancer Research University of California, San Diego

On May 23, 2019, at 5:29 PM, Austin-s-h ***@***.***> wrote: Actually, I followed through the tutorial just a little bit more and found that the UMI cutoff should be much higher, after that, it is 580 barcodes, which is exactly in the range I expected. Thanks! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#12?email_source=notifications&email_token=ABT6GG3V6TOCR52HYFJ6M23PW4EB7A5CNFSM4HPKNOM2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWDRJ6A#issuecomment-495391992>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABT6GGY2UCCFDMQ3WC6EBH3PW4EB7ANCNFSM4HPKNOMQ>.