Splitting BED file by technical replicate or extract all unique barcodes

Question

[summer_researcher]

First post to the forum, sorry I'm new at this:

I inherited a BED file which has 32-length barcodes.

1) The first 8 digits are the cell's unique code

2) The next 8 match an entry from one of three sets (either 7a, 7b or 7c)

3) The next 8 digits are also a part of the cells unique code

4) The last 8 digits match one of two sets (5a or 5b)

We have six technical replicates that are defined as some combination of a 7x and a 5x code.

I did a lot of analysis with this data in Signac/Seurat this summer; I was given a filtered Seurat object to work with. Now, I'm trying to re-do the QC (most importantly, re-do the doublet analysis), and I should input the data as a separate BED file for each of the six technical replicates. This is my first time having access to the original BED file.

I have a lot of the barcodes in my Seurat object but because of filtering it doesn't have all of them. I have a simple script that sorts the barcodes into one of the six technical replicates, but I don't have a list of all the barcodes in the BED file (I'm starting to realize this is usually given separately, but I didn't inherit it).

So I guess my question is whether there is a way to do this filtering entirely in terminal (easily) and/or if there is an easy way to extract all the unique barcodes from a BAM or BED file. [After extracting the barcodes, I know how to make 6 whitelists (.csv) and then split using sinto].

My data looks like this:

chr1    10126   10175   ATTCAGAAAAGAGGCAACGTCCTGTCTTACGC    1
chr1    10150   10180   ATTACTCGAACCAGGTAGGATAGGCTCCTTAC    1
chr1    10151   10192   GAGATTCCCGTATAGAAGGATAGGACTCTAGG    1

[jared.andrews07]

Oh, is it space delimited rather than tab delimited? I also forgot to throw a sort in there. Can you try: cut -f4 -d" " your_file.bed | sort | uniq > output.txt

Using uniq -c will also give the counts for each barcode if that's of interest.

[summer_researcher]

Thanks again, I think its tab delimited because your command re-produced the same BED file. Should I run:

cut -f4 -d"\t" your_file.bed | sort | uniq > output.txt

to indicate tab delimitated?

[jared.andrews07]

cut expects tab delimited by default and your example above works for me when tab delimited. I'd check the file to try to figure out what's going on.

[summer_researcher]

Worked perfectly! Thanks-

[jared.andrews07]

Shouldn't all of these barcodes be unique then?

cut -f4 your_file.bed | uniq > output.bed will get all the unique barcodes if there truly are repeats, but there really shouldn't be unless I'm missing something.

[summer_researcher]

Thanks, I'll give that a try. I think the BED file has a row for each barcode-chromosome position pair. So if I have 10,000 cells and each has an average of 20,000 unique features then doesn't the bed file have 10,000 times 20,000 rows but only 10,000 unique barcodes?

[summer_researcher]

This code made 90 million rows of output but there are only 40,000 cells. Is there a way to tell it that I want unique barcodes (f4) and not unique rows? Thanks again -