SUMMARY: gKaKs is a codon-based genome-level Ka/Ks computation pipeline developed and based on programs from four widely used packages: BLAT, BLASTALL (including bl2seq, formatdb and fastacmd), PAML (including codeml and yn00) and KaKs_Calculator (including 10 substitution rate estimation methods). gKaKs can automatically detect and eliminate frameshift mutations and premature stop codons to compute the substitution rates (Ka, Ks and Ka/Ks) between a well-annotated genome and a non-annotated genome or even a poorly assembled scaffold dataset. It is especially useful for newly sequenced genomes that have not been well annotated. We applied gKaKs to estimate the genome-wide substitution rates in five pairs of closely related species. The average Ka and Ks computed by gKaKs were consistent with previous studies. We also compared the Ka, Ks and Ka/Ks of mouse and rat orthologous protein-coding genes estimated by gKaKs and based on the alignments generated by PAL2NAL. Results from two methods are compatible. AVAILABILITY AND IMPLEMENTATION: gKaKs is implemented in Perl and is freely available on http://longlab.uchicago.edu/?q=gKaKs. The detailed user manual is available on the website.

MOTIVATION: Retrocopies are important genes in the genomes of almost all higher eukaryotes. However, the annotation of such genes is a non-trivial task. Intronless genes have often been considered to be retroposed copies of intron-containing paralogs. Such categorization relies on the implicit premise that alignable regions of the duplicates should be long enough to cover exon-exon junctions of the intron-containing genes, and thus intron loss events can be inferred. Here, we examined the alternative possibility that intronless genes could be generated by partial DNA-based duplication of intron-containing genes in the fruitfly genome. RESULTS: By building pairwise protein-, transcript- and genome-level DNA alignments between intronless genes and their corresponding intron-containing paralogs, we found that alignments do not cover exon-exon junctions in 40% of cases and thus no intron loss could be inferred. For these cases, the candidate parental proteins tend to be partially duplicated, and intergenic sequences or neighboring genes are included in the intronless paralog. Moreover, we observed that it is significantly less likely for these paralogs to show inter-chromosomal duplication and testis-dominant transcription, compared to the remaining 60% of cases with evidence of clear intron loss (retrogenes). These lines of analysis reveal that DNA-based duplication contributes significantly to the 40% of cases of single exon gene duplication. Finally, we performed an analogous survey in the human genome and the result is similar, wherein 34% of the cases do not cover exon-exon junctions. Thus, genome annotation for retrogene identification should discard candidates without clear evidence of intron loss. CONTACT: [email protected]; [email protected]

Abstract Summary: IE-Kb (Intron Exon-Knowledge base) illustrates the intron–exon dynamics in eukaryotic genes. We have developed three different knowledge sets, namely ‘Non-redundant ExInt’, ‘Non-redundant Pfam-ExInt complement’ and ‘Non-redundant GenBank eukaryotic subdivisional sets’ to understand this phenomenon. Statistical analysis is performed on each knowledge set and the results are made available online. The entries in knowledge sets are ranked based on their intron length, exon length and protein length with relational hyper-links to the corresponding intron phase, intron position, intron sequence, gene definition and parent GenBank entry. Availability: http://intron.bic.nus.edu.sg/iekb/iekb.html Contact: [email protected] * To whom correspondence should be addressed.