Skip to main content

Table 3 Effect of read-depth and genotype quality (GQ) thresholds on the accuracy of GBS genotype calls

From: Low-depth genotyping-by-sequencing (GBS) in a bovine population: strategies to maximize the selection of high quality genotypes and the accuracy of imputation

Dataset ID

Filtering criteria (first step)

Number of variants

Markers shared with SNP50

Number of calls examined

Estimated accuracy (%)a

1

≥3 reads

182,334

1150

20,845

92.8

2

≥3 reads and GQ > 20

154,991

1006

11,029

97.4

3

≥4 reads

146,936

956

14,552

95.3

4

≥4 reads and GQ > 20

130,813

865

8873

98.3

5

≥5 reads

122,234

825

9818

97.0

6

≥5 reads and GQ > 20

114,588

778

6952

98.8

7

≥6 reads

105,085

723

6478

97.7

8

≥6 reads and GQ > 20

101,552

689

5147

99.0

  1. aThe reliability of GBS genotypes was based on the reference calls generated using the SNP50 BeadChip. Estimation of the concordant SNPs is reported