Comment on "Evidence of abundant purifying selection in humans for recently acquired regulatory functions"

Science

Volumn 340, Issue 6133, 2013, Pages

  Green, Phil ^a   Ewing, Brent ^a  

^a UNIVERSITY OF WASHINGTON   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

NUCLEOTIDE DERIVATIVE;

GENE FREQUENCY; GENETIC POLYMORPHISM; GENETIC REGULATION; GENETIC VARIABILITY; GENOME; GENOTYPE; HETEROZYGOSITY; HUMAN; MUTATIONAL ANALYSIS; NONHUMAN; NOTE; PRIORITY JOURNAL; PURIFYING SELECTION; SINGLE NUCLEOTIDE POLYMORPHISM; TRANSCRIPTION INITIATION SITE;

EID: 84877601532     PISSN: 00368075     EISSN: 10959203     Source Type: Journal    
DOI: 10.1126/science.1233195     Document Type: Note

References (9)

1. L. D. Ward, M. Kellis, Science 337, 1675 (2012).
2. The ENCODE Project Consortium, Nature 489, 57 (2012).
3. Mouse Genome Sequencing Consortium, Nature 420, 520 (2002).
4. D. R. Bentley et al., Nature 456, 53 (2008).
5. 1000 Genomes Project Consortium, Nature 467, 1061 (2010).
6. We used the 1000 Genomes inferred ancestral allele where available (96% of SNPs), and the major (most common) allele at other SNP sites (the major allele is nonancestral for roughly 15% of SNPs, so using it at 4% of SNPs introduces an error rate of 0.6%).
7. In estimating bases under selection, we modified the authors' linear interpolation procedure (1): For SNP-density- and heterozygosity-based estimates, we used 25 background selection (B) bins rather than 10. For DAF-based estimates, we did not control for B because for small samples, DAF is only weakly correlated with B. We instead controlled for read-depth variation, as follows. For a given region type, we binned SNPs by total read depth in the 59 YRI individuals, using bins of 0 to 9 reads, 10 to 19 reads, and so on; computed the allele frequency distribution for each bin; and took the weighted average of these distributions across bins, using as bin weights the fraction of ENCODE-region SNPs falling into that bin. (Thus, a single set of weights was applied for all region types.) This average distribution was then used to compute the adjusted DAF and adjusted 1+2 (fraction of SNPs with derived allele count 1 or 2) for the interpolation procedure.
8. L. Duret, N. Galtier, Annu. Rev. Genome Hum. G 10, 285 (2009).
9. International HapMap 3 Consortium, Nature 467, 52 (2010).