No intervals loaded

[apritch]

Hi. Thank you for pushing a fix for the install error. I am now running into a new issue:

[M::main] pomfret v0.1-r15
[M::main] CMD: pomfret methphase -o mother --vcf mother_split_sorted.vcf.gz mother_haplotagged_sorted.bam
[M::sancheck_cliopt] read coverage not provided, will estimate.
[M::load_intervals_from_file] at ref chr21

[T::load_intervals_from_file] used 0.1s
[E::blockjoin_parallel] No intervals loaded, terminating.

Any suggestions? I have a trio of ont wgs data. For speed of testing, I have cut this down to chr21. The data has been variant called with clair3, joint genotyped with glnexus, phased with whatshap and i'm now testing improving the phasing with pomfret. The whatshap phase was ran using the joint vcf and the 3 aligned bams. This output a joint phased vcf. I can see PS flags on some but not all variant calls. I then split this vcf to run whatshap haplotag for the three samples (mum, dad, proband). It is the split vcf and haplotagged bam output that is input for this command. Both sorted and indexed. The bam has MD flags. The sample names and nomenclature of 'chr21' match in bam and vcf. Any suggestions? Many thanks in advance.

[apritch]

If useful, I have tested running with the -gtf instead and get the same error as above.

If I run with the vcf and -u (even though the bam should be tagged), the error is:

[dbg::pre_haplotagging_read_in_one_ref] tagged: 28945 new hap0, 28504 new hap1, 29278 new unphased, 0 dup
[T::load_intervals_from_file] used 25.6s
[E::blockjoin_parallel] Nothing loaded from vcf (ref_n=1), cannot haptag the input bam. Terminating.

Thanks again.

[xfengnefx]

Sorry for the late reply, I was away. Could you check whether the variants in the VCF contain phaseblock IDs..? If the PS tag is present but its value is ., it is as if the PS tag did not exist. I believe they should as long as there are phased variants, but I failed to reproduce the error using a phased VCF input. If the VCF could be shared for solely troubleshooting purposes that would be greatly helpful.

The error with --gtf is directly related to the VCF issue. The assumption was that a phased VCF should always have phasblocks. If this is not true for all workflows, I can remove this check, and also add a routine to infer phase blocks from phased variants.