Hi, I have downloaded and aligned PacBio HIFI NA12878 sample against GRCh38 genome, using Minimap2.

java -jar $PICARD CollectWgsMetrics I=NA12878_CCS_sorted.bam O=collect_wgs_metrics.txt R=$ref_fasta

Then I use picard collectwgsmetrics to calculate the coverage, I found that the coverage is 0.0. Then I used samtools flagstat and got this result:

$ samtools flagstate NA12878_CCS_sorted.bam

14410787 + 0 in total (QC-passed reads + QC-failed reads)
4878353 + 0 secondary
450400 + 0 supplementary
0 + 0 duplicates
14396335 + 0 mapped (99.90% : N/A)
0 + 0 paired in sequencing
0 + 0 read1
0 + 0 read2
0 + 0 properly paired (N/A : N/A)
0 + 0 with itself and mate mapped
0 + 0 singletons (N/A : N/A)
0 + 0 with mate mapped to a different chr
0 + 0 with mate mapped to a different chr (mapQ>=5)

It seems like all of the reads are mapped, but how come they have zero coverage? I also checked the alignment manually, and found many read have 0 flag (this one is very subjective).

How did those 2 tools have quite different result? And what is the correct way to calculate coverage of long read sample?