We estimate the penetrance of LQTS for KCNH2 D821G is 20%. We are unaware of any observations of this variant in individuals. D821G is not present in gnomAD. We have tested the trafficking efficiency of this variant, 7% of WT with a standard error of 2%; in our analysis we used SE < 20% as 'high quality'. Approximately below 50% of WT is considered PS3 moderate and below 30% is PS3 strong. D821G has not been functionally characterized. This residue is located in a Mild_Hotspot region for LQT2. In silico predictions, functional data (if available), and location in structure are equivalent to observing 1 individuals with LQT2 and 9 unaffected individuals.These data combined with observations of carriers lead us to estimate the LQTS penetrance for KCNQ1 D821G around 20% (1/10).

PROVEAN	PolyPhen-2	BLAST-PSSM	REVEL	Penetrance Density (%)
-5.861	0.971	-2	0.906	38

PROVEAN scores less than -2 are considered deleterious. REVEL scores higher than 0.5 or 0.75 are considered likely pathogenic (higher sensitivity with the former cutoff, higher specificity with the latter cutoff). A PolyPhen-2 score of 0.85 or greater is considered likely pathogenic. BLAST-PSSM reflects the evolutionary conservation of residue substitutions, more negative numbers indicate fewer observations of the specific substitution than is expected. Penetrance density is our previously published method to calculate the average LQTS probability density in a shell of residues surrounding a residue of interest (Kroncke et al. 2019).

Summary totals might not agree with the literature table because of duplicate patients, which were excluded from the total counts. We do not distinguish here between multiple missense codons. Missense variants are combined across degenerate codon substitutions since codon-level data were not consistently available for curation.

A residue within a folded protein on average has nearest neighbors that fall roughly into two shells: a "nearest" neighbor around 5-6 angstroms and a second shell around 11 angstroms. NOTE: some residues appear multiple times at different distances since the functional K_V11.1 channel (protein product of KCNH2/hERG) is a homotetramer and occasionally the same residue from multiple subunits is present within the 15A window. All variants shown in the rightmost column have been observed in at least one individual in the literature or gnomAD.