Number of barcodes doesn't match with cellranger output.

[NKleinenkuhnen]

Hey,

Thank you very much for your awesome tool! I recently ran into the problem while trying to create a snap file from the output of cellranger. So far I tried to entry points: 1. the position sorted bam file 2. the fragment tsv file. However, in both approaches I ended up with way more barcodes in my snap file than I got in the result report from 10x. In scenario 1 I get 40k barcodes and in scenario 2 20k. According to the 10x summary the dataset should contain 8199 cells. I followed your excellent step-by-step tutorial (https://github.com/r3fang/SnapATAC/wiki/FAQs#10X_snap) and just copied the commands and changed the filenames. I worked with Python 3.7 and the latest version of SnapTools on my Mac. Importing the snap file into R and processing it works like a charm but I couldn't solve the barcode issue myself. I should note that in scenario 1 I had two samples which I processed separately with the same commands and then merged them via createSnap. I hope I could provide you enough information. If you need more just let me know. Thanks in Advance!

[mej54]

Hi there,

I've also noticed a difference between the barcodes based on CellRanger output. I've been using snap-pre with possorted_bam.bam from CellRanger to create snap files as outlined:

cat <( cat $DIR/$SAMPLE.header.sam ) \
<( samtools view $BAM | awk '{for (i=12; i<=NF; ++i) { if ($i ~ "^CB:Z:"){ td[substr($i,1,2)] = substr($i,6,length($i)-5); } }; printf "%s:%s\n", td["CB"], $0 }' ) \
| samtools view -bS - > $DIR/$SAMPLE.snap.bam

samtools sort -n -@ 10 -m 1G $DIR/$SAMPLE.snap.bam -o $DIR/$SAMPLE.snap.nsrt.bam

When I was looking into the promoter ratio using the single_cell.csv files, I noticed there were barcodes in the snap files that were not in the single_cell.csv files (which I believe should contain all fragments). I looked into this further and I'm wondering if at some point snap-pre is taking information from the "CR:Z" flag in the bam instead of the error-corrected barcodes "CR:B"? When I search the barcodes in the snap file that weren't found in the single_cell.csv file, they match to the barcodes under CR:Z, not CR:B (even though the CR:B barcode was added to the read name as outline above).

Thanks!