Adding optional tags to a BAM file

Question

[sc243]

Hi Guys,

I have a question regarding my bam file (single-cell rna seq data). The first 6 letters of the read names of my bam file represent the cellbarcode info and next 6 represent UMI (separated by an underscore and afterwards there is a hash separating it from the rest of the name). For example-

ACAGTG_GAGAAG#K182:79:HV56XX:6:1101:11160:388737da9

I would like to convert these info into the CB CB:Z:ACAGTG and UB UB:Z:GAGAAG bam TAGs. And then the final read name becomes- K182:79:HV56XX:6:1101:11160:388737da9.

Can you please provide me with a one-liner awk command or some kind of tool using which I can do this?

Thanks a lot!

[sc243]

this works perfectly! thanks a lot!

[Pierre Lindenbaum]

update 2021: samtools 1.13: samtools view --add-flags/--remove-flags

[Christina]

Can you provide more information about how the new samtools feature --add-flags works? I would like to be able to use this feature, but for me it doesn't add a tag. For example, I tried to add a CB tag "AAAAAAAAAA" to each read in a small test bam:

samtools view --add-flags 'CB:Z:AAAAAAAAAA' test.bam | head -n 1

Results in the original bam read (i.e., with no new CB tag):

SOLEXA-1GA-2_2_FC20EMB:5:251:979:328 0 chr1 10145 25 36M * 0 0 AACCCCTAACCCTAACCCTAACCCTAACCCTAAACThhhhHcWhhHTghcKA_ONhAAEEBZE?H?CBC?DA NM:i:1 X1:i:1 MD:Z:33A2

[Pierre Lindenbaum]

nice catch. I was wrong --add-flag is about mapping flag (integer), not attributes, my solution above was wrong.

Answer 1

[Pierre Lindenbaum]

using samjdk: http://lindenb.github.io/jvarkit/SamJdk.html

$ cat jeter.sam 
@SQ SN:ref  LN:45
ACAGTG_GAGAAG#K182:79:HV56XX:6:1101:11160:388737da9 1   ref 7   30  8M4I4M1D3M  =   37  39  TTAGATAAAGAGGATACTG *   XX:B:S,12561,2,20,112

.

$ java -jar dist/samjdk.jar -e 'String s=record.getReadName(); int h=s.indexOf("#"),u=s.indexOf("_");record.setReadName(s.substring(h+1));record.setAttribute("CB",s.substring(0,u));record.setAttribute("UB",s.substring(u+1,h));return record;' jeter.sam  2> /dev/null 
@HD VN:1.5  SO:unsorted
@SQ SN:ref  LN:45
K182:79:HV56XX:6:1101:11160:388737da9   1   ref 7   30  8M4I4M1D3M  =   37  39  TTAGATAAAGAGGATACTG *   CB:Z:ACAGTG UB:Z:GAGAAG XX:B:S,12561,2,20,112

[sc243]

Hi Pierre, Just one more question, if the format is rather K182:79:HV56XX:6:1101:11160:388737da9_ACAGTG_GAGAAG can you please tell me the modified samjdk.jar command to provide the tags? Thanks! That will be really helpful!

PS: May be this? Just worked on it....

$ java -jar dist/samjdk.jar -e 'String s=record.getReadName(); int h=s.indexOf("_"),u=s.lastIndexOf("_");record.setReadName(s.substring(0,h-1));record.setAttribute("CB",s.substring(h+1,u));record.setAttribute("UB",s.substring(u+2,u+7));return record;' jeter.sam  2> /dev/null

[Pierre Lindenbaum]

something like (? not tested)

String tokens[] = record.getReadName().split("[_]"); record.setReadName(tokens[0]);record.setAttribute("CB",tokens[1]);record.setAttribute("UB",tokens[2]); return record;

?

Answer 2

[benformatics]

This can be done in R but it's not exactly fast...

library(GenomicAlignments)
library(rtracklayer)

## setting up all bam flags (in my case this a BWA alignment and these are the default cols/flags)
param <- ScanBamParam(what=c("qname","flag","rname", "strand", "pos", "qwidth", "mapq", "cigar","mrnm","mpos","isize","seq","qual"),tag=c("NM","MD","MC","AS","XS"))

## reading original bam file
gr.aln <- readGAlignments("your_bam_file.bam", param=param, use.names=TRUE)

## adding my custom bam tag "RX" (in this case you could use CB/UB etc..)
## in my case the UMI is part of the read name presented as "UM1_UM2_READNAME" you need to modify the regex below for your use case
mcols(gr.aln)$RX <- sub("_.*","",sub("_","",mcols(gr.aln)$qname))

## this would be yours
mcols(gr.aln)$CB <- sub("_.*","",mcols(gr.aln)$qname)
mcols(gr.aln)$UB <- sub(".*_","",sub("#.*","",mcols(gr.aln)$qname))


## export new BAM file with UMI's stored as RX tag
export(gr.aln,'your_bam_file_with_umi.bam')