What does secondary (and some other fields) mean in flagstat output?

Question

[Niek De Klein]

This is my samtools flagstat output for the alignment of a paired-end sample using HISAT (filtered with MQ<1):

45084184 + 0 in total (QC-passed reads + QC-failed reads)
4717987 + 0 secondary
0 + 0 supplementary
0 + 0 duplicates
45084184 + 0 mapped (100.00%:-nan%)
40366197 + 0 paired in sequencing
20273012 + 0 read1
20093185 + 0 read2
39146254 + 0 properly paired (96.98%:-nan%)
39644363 + 0 with itself and mate mapped
721834 + 0 singletons (1.79%:-nan%)
260722 + 0 with mate mapped to a different chr
235361 + 0 with mate mapped to a different chr (mapQ>=5)
(END)

There's quite a few questions about flagstat on Biostars, but I haven't seen the secondary, supplementary or duplicates flag before. Assuming that this is HISAT specific, are is secondary the number of mapped reads that get mapped using the second (non-global) HISAT index? What would supplementary mean? And are duplicates the amount of reads that map non-uniquely?

Answer 1

[Devon Ryan]

These are defined more formally in the SAM specification, though perhaps the wording there isn't great.

Secondary: One of the many places a multimapper can align. Note that multimappers will have one primary and 0 or more such secondary alignments.

Supplementary: For chimeric/fusion/non-linear alignments, this is the location of one part of the alignment. For reads that align in a chimeric fashion, one segment will be designated as primary and the remainder supplementary.

Duplicate: If you've marked possible PCR duplicates, then this will be set. The definition of a duplicate is somewhat dependent on the tool used (N.B., aligners don't typically set this flag, it's down by picard's markDuplicates command or similar).