Predicting Functional Alterations Caused By Non-synonymous Variants in CHO Using Models Based on Phylogenetic Tree and Evolutionary Preservation

Chinese Hamster Ovary (CHO) cell is a major manufacturing platform for one of the most valuable biopharmaceutical products: monoclonal antibodies. Being an immortal cell line adapted to different environments, CHO has been accumulating massive mutations in its genome. Continuous effort has been invested into building a computational model to predict CHO cell productivity. However, not much attention has been focused on its proteins which are surely effected by the mutations accumulated to some extent.
In this project, we focused on the functional effect caused by non-synonymous variants found in CHO genome. A tool was built to firstly identify these variants and then predict their potential function effect by preservation, a concept derived from evolutionary conservation. Firstly, the PANTHER subfamilies, which defined on the base of potential function change within gene trees, were extended by adding proteins from species not covered by PANTHER. Sequences within the same subfamily were then aligned and had Hidden Markov Models (HMMs) built on these alignments. The HMMs were used to identify homologs in CHO proteins. After that preservation were calculated in every site of the alignments, which was then used to predict the function alterations caused by mutations on every site.
Our tool was then validated using data from origin PANTHER subfamilies, PANTHER-PSEP which also calculated site preservation and BLAST, a well-accepted homolog searching algorithm. CHO protein sequences were then imported and analysed by our tool. For comparison, protein sequences from Chinese hamster were also analysed alone with two published CHO cell lines: CHO-K1 and CHO-K1GS. The predictions of proteins from these three genomes were then compared by mapping onto Gene Ontology (GO). Some detailed case studies were also demonstrated. Our tool showed good performance in validations, however, they failed to produce useful hypotheses that would motivate further experiments on bench. The potential causes are discussed at the end.