TABLE 4

Effects of sampling depth on significance of goodness-of-fit test for indel distributions

All indelsDeletionsInsertions
N = 20N = 50N = 100N = 20N = 50N = 100N = 20N = 50N = 100
42.3549.0054.8116.4019.9523.0816.2518.8721.27
(SD)(2.38)(2.48)(2.43)(1.67)(1.62)(1.46)(1.32)(1.36)(1.36)
Tests
>0.05 2.3% 0.8%  0.0% 3.1% 0.5%  0.0% 2.3% 0.8% 0.0%
<0.0597.7%99.2%100.0%96.9%99.5%100.0%64.0%60.8%78.4%
<0.0182.6%90.0% 94.9%84.5%98.8%100.0%22.1%43.6%70.4%
<0.00135.7%50.4% 72.7%78.0%96.5% 99.8%15.4%10.2% 6.7%
<0.000113.2%28.8% 35.9%68.4%88.5% 93.1% 1.3% 0.0% 0.0%
<0.00001 5.1% 3.8%  1.0%47.3%66.7% 84.3% 0.0% 0.0% 0.0%
<0.000001* 0.2% 0.0%  0.0%29.0%52.1% 71.4% 0.0% 0.0% 0.0%
<0.0000001 0.0% 0.0%  0.0%16.9%41.1% 59.9% 0.0% 0.0% 0.0%
  • The number of indels segregating and fraction of significant goodness-of-fit tests from 1000 jackknifing samples of sizes 20, 50, and 100 are shown. The observed number of indels, deletions, and insertions in noncoding regions were 63, 27, and 25, respectively (11 could not be classified as deletions or insertions). Italic numbers correspond to the observed significance level for the whole North American data set (220 alleles) for the respective mutation classes.