Should PP5/BP6 also be removed when the original InterVar PVS1 is 0?

[ChihyiLiao]

Hi AutoGVP team,

Thank you for developing and maintaining AutoGVP. I have a question about how PP5 and BP6 are handled in the InterVar fallback path.

In the workflow figure in the AutoGVP paper, the InterVar branches are marked as including the “removal of PP5/BP6.” I understood this to mean that PP5 and BP6 should not influence classifications produced through the modified InterVar workflow.

However, after reading the current code, I am not sure whether this removal is applied to every InterVar-based classification.

My current understanding is:

Original InterVar PVS1	What AutoGVP appears to do	Possible PP5/BP6 effect
0	Keep the original InterVar classification without recalculating it	If the original classification depended on PP5 or BP6, that effect may remain
1, and AutoPVS1 keeps it as Very Strong	Recalculate the classification	PP5 and BP6 are excluded
1, but AutoPVS1 downgrades or removes PVS1	Still recalculate the classification	PP5 and BP6 are excluded

In other words, the decision to recalculate appears to be based on the original PVS1 value reported by InterVar, before the AutoPVS1 adjustment.

This does not mean that PP5/BP6 are only relevant to variants with PVS1. InterVar can assign PP5 or BP6 to a variant with PVS1=0. The difference seems to be that AutoGVP recalculates variants with original PVS1=1, but directly carries forward the original InterVar classification when original PVS1=0.

I understand that AutoGVP first uses a resolved ClinVar classification when one is available. Therefore, this question only concerns variants that reach the InterVar fallback path, such as variants with:

• no ClinVar record;
• only 0-star ClinVar records; or
• ClinVar records that do not provide a P/LP/VUS/LB/B classification, such as drug-response records.

I am planning to update the ClinVar annotation used by InterVar to the full, unfiltered clinvar_20260530 dataset. While testing this database, I found examples where PP5 or BP6 could change an InterVar classification when the original PVS1 is 0.

PP5 example

• Gene: C2CD3
• GRCh38: chr11:74114384 C>T
• dbSNP: rs755686768
• Transcript: NM_001286577.2:c.1730G>A (p.Arg577His)
• Original PVS1: 0
• Evidence without PP5: PM1 + PM2 + PP3 → VUS
• clinvar_20260530 contains a Likely_pathogenic record with no_assertion_criteria_provided (Allele ID 2753201)
• InterVar can therefore add PP5: PM1 + PM2 + PP3 + PP5 → Likely pathogenic

If the original InterVar classification is copied because PVS1=0, the effect of PP5 appears to be retained.

BP6 example

• Gene: KCNAB2
• GRCh38: chr1:6051658 G>A
• dbSNP: rs530635823
• Transcript: NM_001199862.2:c.122G>A (p.Arg41Gln)
• Original PVS1: 0
• Evidence without BP6: BS1 + PP2 → VUS
• clinvar_20260530 contains a Likely_benign record with no_assertion_criteria_provided (Allele ID 3206907)
• InterVar can therefore add BP6: BS1 + PP2 + BP6 → Likely benign

Again, this matters only when a resolved ClinVar classification is not selected and AutoGVP falls back to InterVar.

The relevant parts of the code appear to be:

• PP5 and BP6 are excluded from the evidence counts used for recalculation
• The recalculation branch is selected using the original InterVar PVS1 value
• For other variants, the original InterVar classification is retained

Could you please confirm whether my reading of the workflow is correct?

More specifically:

1. Is it intentional that the possible effects of PP5/BP6 are retained when the original InterVar PVS1 is 0?
2. If PP5/BP6 are intended to be removed from all InterVar fallback classifications, should the PVS1=0 variants also be recalculated without these two criteria?
3. Alternatively, are users expected to filter records such as no_assertion_criteria_provided from the ClinVar database before using it with InterVar?
4. Should the same PP5/BP6 exclusion also be applied when update_intervar.R recalculates classifications after updating PS1 or PM5?

I may be missing another step in the intended workflow, so I would appreciate any clarification before changing our local pipeline.

Thank you!