Document criteria filled out by InterVar and AutoCNV

[holtgrewe]

Is your feature request related to a problem? Please describe.
We currently do not explain well the ACMG criteria filled out by InterVar and AutoCNV.

Describe the solution you'd like
We should both summarize the criteria and refer to the InterVar and AutoCNV publications.

Describe alternatives you've considered
N/A

Additional context
N/A

[xiamaz]

Papers:

Intervar - https://www.cell.com/ajhg/fulltext/S0002-9297(17)30004-6

Autocnv - https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-021-08011-4

[xiamaz]

Intervar Criteria

see Methods > Criteria and Scoring System

Automated criteria

Criteria	Rule
PVS1	Null variants on canonical transcript for 4807 identified LOF-intolerant gene list, before 50 nucleotides of final exon-junction complex
PS1	Automatic match against list of ClinVar pathogenic missense-variants, same AA change
PS4	Variants with OR > 5.0 in GWASdb v2
PM1	domain info from dbnsfp31a_interpro database, list of domains with only pathogenic and likely pathogenic variants based on ClinVar data
PM2	absent in ESP6500, 1000 Genomes, ExAC for dominant or AAF <0,5% for recessive
PM4	non-frameshift insertion/deletion, stop-loss in non-repeat regions (rmsk database UCSC browser)
PM5	Automatic match against list of ClinVar pathogenic missense-variants, different AA change
PP2	>80% pathogenic (at least one) clinvar variants missense and <10% benign (and less than one)
PP3	dfnsfp30a MetaSVM (>0, deleteriousness), GERP++ (>2.0, conservation), dbscnv11 (>0.6 ADA, RF scores)
PP5	ClinVar or HGMD as database
BA1	AAF > 5%
BS1	AAF > 1% (default cutoff, user-adjustable)
BS2	hom (for AR) or het (for AD) in 1000 Genomes
BP1	>80% pathogenic (at least one) clinvar variants truncating
BP3	non-frameshift insertion, non-frameshift deletion in repeat region (defined by rmsk database)
BP4	Evidence (see PP3) does not suggest impact
BP6	ClinVar or HGMD as database
BP7	dbscnv RF and ADA <0.6, GERP++ <2 (not conserved)

Non automatically scored criteria

PS2, PM6 - de novo status of variant
PS3, BS6 - functional studies
PM3, BP2 - variant in cis/trans with known pathogenic
PP1, BS4 - familial segregation
PP4 - phenotype and family history
BP5 - alternative molecular basis

[xiamaz]

AutoCNV

Automated scorings

Section	Evidence	Rule
1	1A	0 otherwise
	1B	-0.6 if no protein coding genes or functionally important elements
2	2A	1 if del spans haploinssuficient or dup spans triplosensitive gene or region (clingen database)
	2B	0 no overlap
	2C	0.9 if exon involved or 0.0 without for partial 5' overlap
	2D	0.9 for 3' overlap if pathogenic variants documented in exon (P/LP ClinVar /w AF <1% gnomAD), 0.3 without known pathogenic, 0.9 multiple exons
	2E	AutoPVS1, 0.9 if NMD, 0.45 if altered region critical, 0.45 if >10% protein removed, 0.3 <10% protein
	2F	-1 if established benign genes, regions
	2G	0 if established benign genes but includes additional regions
	2H	0.15 gene pLI >=0.9 and Decipher HI <=10%
	2I	dups AutoPVS1, 0.9 if tandem + NMD, 0.45 if tandem
	2L	dups 0 genes without clinical significance
3	3A	0 otherwise
	3B	0.45 del 25-34 or dup 35-49 protein coding genes
	3C	0.9 for del >35 and dup >50 protein coding genes
4	4O	-1 if CNV entirely within common variation (DGV Freq >=1% or gnomAD >=1%), or if overlap >50% without containing other protein coding genes

Non automatically scored criteria

Section 2 - 2J, 2K - patient phenotype consistency with LOF of gene
Section 4 - 4A-N - phenotype, segregation in literature, case control
Section 5 - 5A-5H - patient phenotype, family segregation

[xiamaz]

@holtgrewe @gromdimon @sczakihl I have added a table and written a brief text in the comparison section of the paper. Just realized that the CNV rule numbering greatly differs between dels and dups. So the current table is a bit incomplete.

[holtgrewe]

@xiamaz will you update the table here as well?

[xiamaz]

@holtgrewe Which table do you refer to?

[holtgrewe]

So the current table is a bit incomplete.

@xiamaz this one ;-)

[xiamaz]

@holtgrewe Only if you actually need that info in a figure, as the AutoCNV paper is not very clear on some details.

[holtgrewe]

@xiamaz can you take over #440 and fill out the missing information and review the existing one