{readname} did not have a primary R1 record. (GroupReadsByUmi)

[xiechangxiao]

when I use GroupReadsByUmi group reads, this error occurs.@nh13
cmd:
java -jar fgbio-2.2.1.jar --tmp-dir=/tmp GroupReadsByUmi --input=test.bam --output=test.umi.group.BAM --strategy=paired --min-map-q=20 --edits=1 --raw-tag=RX --family-size-histogram=test.family.size.histogram

error message:
Exception in thread "main" java.lang.IllegalStateException: {readname} did not have a primary R1 record

[nh13]

Can you tell us how you created the test bam, and the provide a test case (BAM) to debug? Did you sort the reads prior?

[xiechangxiao]

First, mapping with BWA:
bwa mem -M -K 10000000 -R "@RG\tID:1\tPL:Illumina\tLB:hg19\tSM:tumor" hg19.fasta Tumor_1.clean.fq.gz Tumor_2.clean.fq.gz |samtools view -b -o Tumor.raw.bam - && samtools sort Tumor.raw.bam -o Tumor.sort.bam -O bam
Then, extract MQ and RX:
samtools view -h Tumor.sort.bam | mawk '/^@/ {print;next} {N=split($1,n,":");gsub("_", "-", n[N]);print $0 "\tMQ:i:" $5 "\tRX:Z:" n[N]}' |samtools view -b - >test.bam
Finally, use GroupReadsByUmi and get errors:
Exception in thread "main" java.lang.IllegalStateException: E250006338L1C038R0044055987:GTGAG_GCACG did not have a primary R1 record.
But, the flag value of this paired read is 113 and 177.@nh13

[nh13]

Can you confirm there is a R1 primary record in your test.bam "E250006338L1C038R0044055987:GTGAG_GCACG"? Either your RX/MQ extraction isn't working as intended, or your sort order into GroupReadsByUmi is wrong (try samtools sort --template-coordinate using the latest samtools). Having test.bam would be helpful to debug futher.

To extract the UMI bases, try FastqToBam or other tools in fgbio, which we support, and are used in our best-practices pipeline:

https://github.com/fulcrumgenomics/fgbio/blob/main/docs/best-practice-consensus-pipeline.md

[xiechangxiao]

I provide a simple bam file for you debug.@nh13
test.zip
If you remove "E250006338L1C038R0044055987:GTGAG_GCACG" from this bam file, GroupReadsByUmi will work properly.
By the way, I've followed your best practices, but it's time consuming and space consuming.
Personally, I think there could be easier way to build a consensus pipeline.
For example:

1. shift UMI to readname with fastp
2. mapping with BWA
3. GroupReadsByUmi and CallMolecularConsensusReads