A large number of non-synonymous single-nucleotide polymorphisms (nsSNPs) have been found in human genome, but there is poor knowledge on the relationship between the genotype and phenotype of these nsSNPs. Human ATP-binding cassette (ABC) transporters are able to transport a number of important substrates including endogenous and exogenous compounds. This study aimed to predict the phenotypical impact of nsSNPs of human ABC transporter genes, and the predicted results were further validated by reported phenotypical data from site-directed mutagenesis and clinical genetic studies. One thousand and six hundred thirty-two nsSNPs were found from 49 human ABC transporter genes. Using the PolyPhen and SIFT algorithms, 41.8-53.6% of nsSNPs in ABC transporter genes were predicted to have an impact on protein function. The prediction accuracy was up to 63-85% when compared with known phenotypical data from in vivo and in vitro studies. There was a significant concordance between the prediction results using SIFT and PolyPhen. Of nsSNPs predicted as deleterious, the prediction scores by SIFT and PolyPhen were significantly related to the number of nsSNPs with known phenotypes confirmed by experimental and human studies. The amino acid substitution variants are supposed to be the pathogenetic basis of increased susceptibility to certain diseases with Mendelian or complex inheritance, altered drug resistance and altered drug clearance and response. Predicting the phenotypic consequence of nsSNPs using computational algorithms may provide a better understanding of genetic differences in susceptibility to diseases and drug response. The prediction of nsSNPs in human ABC transporter genes would be useful hints for further genotype-phenotype studies.

The nuclear receptor (NR) superfamily represents an important group of regulating factors that control the expression of a number of target genes including those encoding important drug metabolizing enzymes and drug transporters. Single nucleotide polymorphism (SNP) is the most common mutation in the human genome and a large number of SNPs have been identified to date. It is unlikely to examine the functional impact of all these mutations using an experimental approach. As such, we employed two algorithms, Sorting Intolerant from Tolerant (SIFT) and Polymorphism Phenotyping (PolyPhen) to predict the impact of non-synonymous SNPs (nsSNPs) on NR activities and disease susceptibility. We identified 442 nsSNPs in a systematic screening of 48 human NR genes. Using SIFT, of 442 amino acid substitutions, 289 (65.38%) were classified as "intolerant". The PolyPhen program classified 269 (60.86%) of them as "probably damaging" or "possibly damaging". The results from the two algorithms were in concordance. Among the 442 mutations, 229 of them have been functionally characterized. SIFT predicted 192 of these nsSNPs as "intolerant", resulting in a correct prediction rate of 83.84%, while PolyPhen gave a prediction rate of 76.86%. For 216 nsSNPs of the androgen receptor gene, 149 nsSNPs have been functionally studied and most (121) of them resulted in a reduction of receptor activity. SIFT sorted 187 out of 216 as "intolerant" (86.57%) and PolyPhen identified 159 out of 216 as "potentially intolerant" (73.61%). These results indicate that both SIFT and PolyPhen are useful and efficient tools to predict the functional effects of nsSNPs of human NR genes.