Changing References

Some vg formats, like GBZ, distinguish between paths of different types, or "senses".

For example, we can put a P-line path chr1 and a haplotype for sample sample on contig chr1 into a GBZ file:

cat >demo.gfa <<EOF
H	VN:Z:1.1
S	1	GATTACT
S	2	A
S	3	T
S	4	CATTAG
L	1	+	2	+	*
L	1	+	3	+	*
L	2	+	4	+	*
L	3	+	4	+	*
P	chr1	1+,2+,4+	*
W	sample1	0	chr1	0	14	>1>3>4
W	sample2	0	chr1	0	14	>1>2>4
EOF
vg gbwt -G demo.gfa --gbz-format -g demo.gbz
rm demo.gfa

Then we can inspect the path metadata as a TSV with vg paths -M:

vg paths -M -x demo.gbz

The result will be:

#NAME	SENSE	SAMPLE	HAPLOTYPE	LOCUS	PHASE_BLOCK	SUBRANGE
chr1	SENSE_GENERIC	NO_SAMPLE_NAME	NO_HAPLOTYPE	chr1	NO_PHASE_BLOCK	NO_SUBRANGE
sample1#0#chr1#0	SENSE_HAPLOTYPE	sample1	0	chr1	0	NO_SUBRANGE
sample2#0#chr1#0	SENSE_HAPLOTYPE	sample2	0	chr1	0	NO_SUBRANGE

Tools like vg surject won't operate on haplotype paths by default. If we want to change the graph to make a haplotype path into a reference path, we can apply a transformation when converting formats to promote haplotypes for a sample:

vg convert -a --ref-sample sample1 demo.gbz >demo-promoted.vg

If we check the path metadata again:

vg paths -M -x demo-promoted.vg

The result will be:

#NAME	SENSE	SAMPLE	HAPLOTYPE	LOCUS	PHASE_BLOCK	SUBRANGE
chr1	SENSE_GENERIC	NO_SAMPLE_NAME	NO_HAPLOTYPE	chr1	NO_PHASE_BLOCK	NO_SUBRANGE
sample1#chr1	SENSE_REFERENCE	sample1	NO_HAPLOTYPE	chr1	NO_PHASE_BLOCK	NO_SUBRANGE
sample2#0#chr1#0	SENSE_HAPLOTYPE	sample2	0	chr1	0	NO_SUBRANGE

Note that the sense of the path for sample sample1 has changed to reference, and that its haplotype and phase block information have been removed. (It is safe to remove the haplotype because it is 0, the value used for a haploid haplotype. If we were using a diploid sample, 1 and 2 would have been used for the haplotypes, and they would have been preserved.)

If we decide we don't want the old original chr1 path in there anymore, we can remove all paths with a chr1 prefix:

vg paths -x demo-promoted.vg -d -Q chr1 >demo-removed.vg

Then we can list the path names:

vg paths -L -x demo-removed.vg

And we will see that it is no longer there:

sample1#chr1
sample2#0#chr1#0

Converting the resulting graph back to GBZ format, for use with vg giraffe, is not yet implemented. However, the modified graph can be used with alignments against the original graph, since the node IDs are all the same.

To clean up:

rm demo.gbz demo-promoted.vg demo-removed.vg