Supplemental Materials for:

 

“Partition-Ligation EM Algorithm for Haplotype Inference with Single Nucleotide Polymorphisms”

 

Legend to Figure A1.

 

Histograms of the average error rates of PL-EM for three different buffer sizes (small, size=20; medium, size=50; large, size=80; the default atomistic unit size of 5-8 was used) based on individual phasing results (open bars) or the proportion of incorrectly inferred loci (shaded bars) for (A) Angiotensin I Converting Enzyme (ACE) data, (B) b2-Adrenergic Receptor (b2AR) data, and (C) Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) data (see Niu et al. 2002). The data sets were generated using the same procedure as described in the legend to Figure 1. The error bars are shown as ± 1 standard error.

 

Legend to Figure A2.

 

Histograms of the average error rates of PL-EM for three different atomistic unit sizes (small: K=3-4; medium: K=5-8; large: K=9-16; the default buffer size of 50 was used) based on individual phasing results (open bars) or the proportion of incorrectly inferred loci (shaded bars) for (A) Angiotensin I Converting Enzyme (ACE) data, (B) b2-Adrenergic Receptor (b2AR) data, and (C) Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) data (see Niu et al. 2002). The data sets were generated using the same procedure as described in the legend to Figure 1. The error bars are shown as ± 1 standard error.

 

Legend to Figure A3.

 

Histograms of the number of times (out of 100 trials) for each individual to be phased incorrectly by the three algorithms: PHASE (blue bar), HAPLOTYPER (red bar), PL-EM (yellow bar; the default atomistic unit size of 5-8 and the default buffer size of 50 were used) for (A) Angiotensin I Converting Enzyme (ACE) data, (B) b2-Adrenergic Receptor (b2AR) data, and (C) Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) data (see Niu et al. 2002). The data sets for (A) were generated using the same procedure as described in the legend to Figure 1. For (B) and (C), only one randomly generated dataset was used throughout the 100 trials. The error bars are shown as ± 1 standard error. The old version of PHASE (Stephens et al. 2001) was applied to (A) due to its better performance over the new version.

 

For academic and non-profit users, you may download PL-EM executable free by click here. If you are working for an for-profit institution, please contact  Ms. Holly Foskett at Harvard Office for Technology and Trademark Licensing, phone: (617) 496-0474, email: holly_foskett@harvard.edu. Be sure to specify whether you are interested in obtaining an executable file or the source code of PL-EM.