Table 4 MLR classification accuracy.

From: Differentiation of Hispanic biogeographic ancestry with 80 ancestry informative markers

SNP Panel

Dataset

HUR

DOM

COL

CUB

PUR

PEL

MXL

Overall

Setser80

GOAL

100% (±0%)

100% (±0%)

100% (±0%)

97.5% (±5%)

97.5% (±5%)

N/A

N/A

99%

Seldin96

GOAL

97.5% (±5%)

95% (±5.8%)

85% (±12.9%)

90% (±11.5%)

95% (±5.8%)

N/A

N/A

92.5%

Kidd44

GOAL

92.5% (±9.6%)

90% (±0%)

75% (±17.3%)

72.5% (±15%)

72.5% (±9.6%)

N/A

N/A

80.5%

Setser80

1000 G

N/A

N/A

90.4% (±7.4%)

N/A

90.4% (±7.4%)

100% (±0%)

94.2% (±7.4%)

93.8%

Seldin96

1000 G

N/A

N/A

94.2% (±3.8%)

N/A

88.5% (±7.7%)

100% (±0%)

92.3% (±6.3%)

93.8%

Kidd44

1000 G

N/A

N/A

76.9% (±8.9%)

N/A

76.9% (±6.3%)

92.3% (±8.9%)

82.7% (±9.7%)

82.2%

Setser80

7 Pops

95% (±5.8%)

97.5% (±5%)

77.5% (±9.6%)

100% (±0%)

92.5% (±9.6%)

100% (±0%)

90% (±8.2%)

93.2%

Seldin96

7 Pops

100% (±0%)

82.5% (±20.6%)

82.5% (±12.6%)

85% (±17.3%)

67.5% (±17.1%)

97.5% (±5%)

100% (±0%)

87.9%

Kidd44

7 Pops

57.5% (±9.6%)

85% (±12.9%)

55% (±5.8%)

72.5% (±12.6%)

55% (±12.9%)

92.5% (±9.6%)

82.5% (±9.6%)

71.4%

  1. Comparison of the nine possible combinations of each of three simulated datasets on each of three SNP panels and their MLR classification accuracy for each population. Reported as percent accuracy with two-tailed standard deviations listed in parentheses (). Abbreviations used: GOAL = Genomic Origins and Admixture in Latinos, 1000 G = 1000 Genomes Project, 7 Pops = 7 Populations Combined, COL = Colombia, CUB = Cuba, DOM = Dominican Republic, HUR = Honduras, PUR = Puerto Rico, PEL = Peru from Lima, MXL = Mexicans living in Los Angeles, and MLR = multinomial logistic regression. Both Colombian populations from GOAL and 1000 G are listed in this table as “COL”.