id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
10_1101-2021_01_06_425550	Chen, Nae-Chyun	Improving variant calling using population data and deep learning	2021	17	.pdf	application/pdf	6912	754	55	We further evaluated the performance of the models using two whole-exome sequencing (WES) datasets from a recently released set of genome and exome data [23] (Figure 2). Among the populationresolved false-positive errors, more than two third (71.0%) are uncommon (allele frequency ≤ 5%) among the 1000Genomes samples, whereas there are only 11.4% uncommon variants for population-induced false positives. This observation supports the hypothesis that the population-aware model uses allele frequency to adjust its variant calls. A potential concern for population-aware variant calling models is increasing false negative rate for novel alleles. To better understand the zero-frequency variants, we called variants using the DeepVariant PacBio model with the PrecisionFDA v2 35x HG003 reads set sequenced with the We evaluate potential biases introduced by population information in variant calling by comparing population-aware models that use allele frequencies from different Despite greater overall accuracy, we note that the population-aware model underperforms on variants with zero allele frequencies in 1000Genomes.	./cache/10_1101-2021_01_06_425550.pdf	./txt/10_1101-2021_01_06_425550.txt