Reference flow: reducing reference bias using multiple population genomes

Table 2 Measures of allelic balance for NA12878 whole genome sequencing dataset stratified by Genome-in-a-Bottle v3.3.2 confidence annotation

Method	REF-to-ALT	Total	# biased	# biased
	ratio	# biased	toward REF	toward ALT
High confidence
GRCh38	1.0407	20,012	18,141	1871
Major	1.0227	19,837	15,415	4422
RandFlow-LD	1.0133	12,239	9512	2727
vg	1.0124	10,518	7971	2547
RandFlow-LD-26	1.0098	9984	7489	2495
Personalized	1.0033	7024	4,600	2424
Low confidence
GRCh38	1.2355	24,798	20,594	4204
Major	1.1230	26,579	20,108	6471
RandFlow-LD	1.0282	22,190	15,891	6299
vg	1.0120	21,266	15,422	5844
RandFlow-LD-26	1.0008	20,333	13,874	6459
Personalized	0.9750	16,266	9299	6967
All regions
GRCh38	1.0718	44,810	38,735	6075
Major	1.0397	46,416	35,523	10,893
RandFlow-LD	1.0160	34,429	25,403	9026
vg	1.0123	31,784	23,393	8391
RandFlow-LD-26	1.0081	30,317	21,363	8954
Personalized	0.9981	23,290	13,899	9391

vg index includes variants with 10% or higher allele frequency in the 1000-Genomes Project GRCh38 call set. The methods are sorted by REF-to-ALT ratio in all regions

ISSN: 1474-760X