GENIE - Generation of count data

---

Count data are generated to assist with variant interpretation according to the UK Somatic Variant Interpretation Guidelines (S-VIG).

Data pre-processing

GENIE data are provided for download in MAF format (GRCh37) against specific MSKCC transcripts. To annotate against MANE transcripts as priority, the variants were processed as follows:

1. Converted to VCF format
   1. During this process variants with mismatched reference bases according to GRCh37 were excluded.
2. Normalised (GRCh37) using bcftools v1.23
   1. Variants within the GENIE dataset are provided by separate submitting institutions and no normalisation is performed, therefore variants were normalised (left-aligned and made parsimonious) to ensure each unique variant was represented in the same way across institutions.
3. Lifted over from GRCh37 to GRCh38 using Picard LiftoverVcf v3.3.0
   1. Variants which failed to lift over were removed.
4. Re-annotated in GRCh38 using the Ensembl Variant Effect Predictor (VEP) v115.2
   1. The --pick parameter was used to report one block of annotation per variant; this prioritises annotation against MANE transcripts using the order described here.
   2. Annotations for Hugo_Symbol, Consequence, RefSeq, HGVSc, HGVSp, and Protein_position were replaced with the annotations from VEP v115.2 in GRCh38 prior to counting.

Generating patient-level counts

Count Type	Present for	Description
Same nucleotide change	All variants.	Variants are grouped by GRCh38 CHROM, POS, REF and ALT and the number of unique patients are counted.
Same amino acid change	All variants with HGVSp notation present.	Variants are grouped by Hugo_Symbol, RefSeq and HGVSp and the number of unique patients are counted.
Same or downstream frameshift (truncating)	Any variants with a Consequence of frameshift_variant or stop_gained which have "Ter" (and not "ext") in the HGVSp notation.	The first position from the Protein_position is extracted and the number of unique patients with downstream frameshift (truncating) variants for the same Hugo_Symbol and RefSeq are counted.
Nested inframe deletions	Any variants with a Consequence equal to inframe_deletion with HGVSp notation present.	For each variant, the deletion range is extracted from the Protein_position and the number of unique patients with an inframe deletion with the same Hugo_Symbol and RefSeq which affects the same range or is nested within the range are counted.

Counting principles

A variant may be present within GENIE data multiple times for the same patient, which can be for the same or different cancer types. If a patient had two tests performed for two different cancer types:

1. Ovarian Cancer
2. Leukemia

The patient would contribute towards the counts for both cancer 1 (Ovarian Cancer) and cancer 2 (Leukemia). However, counts for all cancers (and grouped counts for haemonc or solid cancers) would not double count any variants for this patient shared between cancer 1 & 2.

Examples

Missense variants

Rows in re-annotated GENIE data for two different variants with the same HGVSp:

Consequence	HGVSc	HGVSp	PATIENT ID	CANCER TYPE
Missense variant	NM_000546.6:c.807C>G	NP_000537.3:p.Ser269Arg	GENIE-DFCI-325916	Colorectal Cancer
Missense variant	NM_000546.6:c.807C>G	NP_000537.3:p.Ser269Arg	GENIE-DFCI-617112	Pancreatic Cancer
Missense variant	NM_000546.6:c.805A>C	NP_000537.3:p.Ser269Arg	GENIE-MSK-P-0091153	Mature B-Cell Neoplasms
Missense variant	NM_000546.6:c.805A>C	NP_000537.3:p.Ser269Arg	GENIE-UHN-689198	Non-Small Cell Lung Cancer

Counts generated:

Count type	Cancer grouping	c.807C>G	c.805A>C
Same Nucleotide Change	All Cancers	2	2
	Haemonc Cancers	-	1
	Solid Cancers	2	1
	Pancreatic Cancer	1	-
	Colorectal Cancer	1	-
	Mature B Cell Neoplasms	-	1
	Non-Small Cell Lung Cancer	-	1
Same Amino Acid Change	All Cancers	4	4
	Haemonc Cancers	1	1
	Solid Cancers	3	3
	Pancreatic Cancer	1	1
	Mature B Cell Neoplasms	1	1
	Non-Small Cell Lung Cancer	1	1
	Colorectal Cancer	1	1

Frameshift (truncating) variants

Rows in re-annotated GENIE data for three different variants:

Consequence	HGVSc	HGVSp	Protein position	PATIENT ID	CANCER TYPE
Frameshift variant	NM_001282717.2: c.3052dup	NP_001269646.1: p.Arg1018ProfsTer62	1018	GENIE-COLU-00170	Soft Tissue Sarcoma
Frameshift variant	NM_001282717.2: c.3052dup	NP_001269646.1: p.Arg1018ProfsTer62	1018	GENIE-COLU-1828	Soft Tissue Sarcoma
Frameshift variant	NM_001282717.2: c.1728del	NP_001269646.1: p.Thr577LeufsTer58	576	GENIE-COLU-2025	Soft Tissue Sarcoma
Frameshift variant	NM_001282717.2: c.1435_1436insG	NP_001269646.1: p.Gln479ArgfsTer14	479	GENIE-COLU-2190	Breast Cancer

Counts generated (based on Protein_position):

Count type	Cancer grouping	c.3052dup	c.1728del	c.1435_1436insG
Same Nucleotide Change	All Cancers	2	1	1
	Solid Cancers	2	1	1
	Soft Tissue Sarcoma	2	1	-
	Breast Cancer	-	-	1
Same Amino Acid Change	All Cancers	2	1	1
	Solid Cancers	2	1	1
	Soft Tissue Sarcoma	2	1	-
	Breast Cancer	-	-	1
Same Or Downstream Truncating Variants	All Cancers	2	3	4
	Solid Cancers	2	3	4
	Soft Tissue Sarcoma	2	3	3
	Breast Cancer	-	-	1

Inframe deletions

Rows in re-annotated GENIE data for three different variants with some nesting (based on Protein_position):

Consequence	HGVSc	HGVSp	Protein position	PATIENT ID	CANCER TYPE
Inframe deletion	NM_003722.5: c.567_569del	NP_003713.3: p.Ala190del	189-190	GENIE-MSK-P-0104506	Non-Small Cell Lung Cancer
Inframe deletion	NM_003722.5: c.563_568del	NP_003713.3: p.Lys188_Ala190delinsThr	188-190	GENIE-MSK-P-0029775	Non-Small Cell Lung Cancer
Inframe deletion	NM_003722.5: c.563_571del	NP_003713.3: p.Lys188_Ala190del	188-191	GENIE-MSK-P-0109964	Vaginal Cancer

Counts generated:

Count type	Cancer grouping	c.567_569del	c.563_568del	c.563_571del
Same Nucleotide Change	All Cancers	1	1	1
	Solid Cancers	1	1	1
	Non-Small Cell Lung Cancer	1	1	-
	Vaginal Cancer	-	-	1
Same Amino Acid Change	All Cancers	1	1	1
	Solid Cancers	1	1	1
	Non-Small Cell Lung Cancer	1	1	-
	Vaginal Cancer	-	-	1
Nested Inframe Deletions	All Cancers	1	2	3
	Solid Cancers	1	2	3
	Non-Small Cell Lung Cancer	1	2	2
	Vaginal Cancer	-	-	1

Variants removed

• Variants which were not annotated against a gene (empty gene symbol) were removed.
• Variants with a mismatch against the GRCh37 reference when converting from MAF to VCF description were removed.
• Variants which failed to liftover from GRCh37 -> GRCh38 were removed.
• Variants on alt contigs were removed.