Mark PCR duplicates with samtools markdup

Flag PCR duplicates in an aligned BAM using the four-step samtools collate, fixmate -m, sort, and markdup pipeline.

SDSomenath DuttaUpdated July 21, 2026

Prerequisites

samtools 1.10 or newer installed (`samtools markdup` available)
An aligned BAM file (e.g. sorted.bam)

1Group mates together and add the ms tag

markdup needs each read's mate beside it plus an ms (mate-score) tag, and neither exists in a plain coordinate-sorted BAM. Run collate to pull read pairs together, then fixmate with -m to write the ms and MC tags. Both commands finish silently on success.

bash

samtools collate -@ 4 -o collated.bam sorted.bam
samtools fixmate -m collated.bam fixmate.bam

Expected output

(collate and fixmate print nothing on success; collated.bam and fixmate.bam are written)

2Coordinate-sort, then mark duplicates

markdup requires coordinate order, so sort the fixmate output before marking. Running markdup then flags duplicate read pairs in place using the 0x400 flag rather than deleting them, so the reads stay in the BAM for downstream tools to filter or keep.

bash

samtools sort -@ 4 -o positionsort.bam fixmate.bam
samtools markdup positionsort.bam markdup.bam

Expected output

[bam_sort_core] merging from 4 files and 4 in-memory blocks...

3Verify the duplicate count

Confirm the run with flagstat and filter to the duplicates line. The count reports how many reads now carry the duplicate flag.

bash

samtools flagstat markdup.bam | grep duplicates

Expected output

18342 + 0 duplicates

Troubleshooting

samtools markdup: error, no MC or ms tag

You skipped `samtools fixmate -m`. markdup relies on the MC and ms tags, so always collate first, then run fixmate with -m, before sorting and marking.

You want duplicates removed from the BAM, not just flagged

Add -r to markdup so it drops duplicate reads instead of tagging them: `samtools markdup -r positionsort.bam dedup.bam`.

References

samtools markdup documentation