Prediction of Readthroughs Based on the Statistical Analysis of Nucleotides Around Stop Codons

摘要

Readthrough is an unusual translational event in which a stop codon is skipped or misread as a sense codon. Translation then continues past the stop codon and results in an extended protein product. Reliable prediction of readthroughs is not easy since readthrough is in competition with standard decoding and readthroughs occur only at a tiny fraction of stop codons in the genome. We developed a program that predicts readthrough sites directly from statistical analysis of nucleotides surrounding all stop codons in genomic sequences. Experimental results of the program on 86 genome sequences showed that 80% and 100% of the actual readthrough sites were found in the top 3% and 10% prediction scores, respectively.