Skip to main content

Table 1 Comparison of four algorithms (AWclust, STRUCTURE, BAPS and ipPCA) for inferring the number of primal subpopulations (K).

From: Iterative pruning PCA improves resolution of highly structured populations

Dataset

AWclust

STRUCTURE

BAPS

ipPCA

model 1 simulated (3 subpopulations)

3

3

2

3

model 2 simulated (5 subpopulations)

5

6

5

5

model 3 simulated (20 subpopulations, 30 datasets)

N/A1

N/D2

113

203

HapMap (4 population labels with 50000 SNPs)

3

3

3

4

bovine (9 population labels with 9329 SNPs)

N/A1

10

14

9

Shriver's (14 population labels with 11555 SNPs)

N/A1

144

5

12

  1. 1 Not applicable since AWclust gap statistics limits K < 8
  2. 2 Not done owing to the computational constraint
  3. 3 Modal value from 30 simulated datasets
  4. 4 There are two subpopulations with no individuals assigned to them