We developed a computational framework to make ortholog annotation between closely related species, through whole genome alignment and local sequence alignment followed by multiple filters. The annotation is designed for cross-species transcriptome comparisons using RNA sequencing data, but can also be applied to gene prediction for species with incomplete annotation. Here we tested the previous methods in comparing brain transcriptomes of human and closely related non-human primates (chimpanzee and rhesus macaque), our approach expands ortholog annotation and reduces the false discovery of differentially expressed genes, while keeping the false negative rate low. Additional data and software updates will be incorporated into the database as it becomes available