Skip to main content

Table 1 The performance of ARCHes and RFMix on various test sets

From: Ancestry inference using reference labeled clusters of haplotypes

Test set group

Test set

Global concordance, average over test sets (proportion of test sets with superior performance)

Local concordance, average over test sets (proportion of test sets with superior performance)

ARCHes

RFMix

ARCHes

RFMix

1000 Genomes and HGDP

Single-Origin Testset Examples

66.1% (13/15)

43.5% (2/15)

66.1% (13/15)

43.5% (2/15)

Simulated admixture

45% Native American, 50% European, 5% African

72.3% (1/1)

65.7% (0/1)

47.8% (1/1)

18.5% (0/1)

Simulated admixture from 16 Pairs of neighboring regions

50%-50% admixed (2 simulation founders)

60.1% (11/16)

48.9% (5/16)

58.8% (14/16)

41.8% (2/16)

Approx. 25%-75% Admixed (4)

63.6% (13/16)

51.8% (3/16)

60.1% (14/16)

44.7% (2/16)

Approx. 12.5%-87.5% Admixed (8)

65.2% (13/16)

51.2% (3/16)

62.6% (14/16)

46.0% (2/16)

Approx. 6.25%-93.75% Admixed (16)

66.2% (14/16)

50.0% (2/16)

64.5% (14/16)

47.0% (2/16)

  1. Global concordance is the intersection between the estimated amounts for each region and the amount present in a test example, and local concordance is the number of correct assignments to each genomic window. For single-origin test examples, these measures are the same