Animal models and in vitro cell culture experiments are instrumental to substitute for human studies and expedite basic and clinical research. There have been constant debates of whether non-human models can represent human well but no adequate bioinformatic tool has been developed to adequately answer the question. We propose the following aims in this proposal to provide an answer in the transcriptomic context: (1) We will develop a transcriptomic resemblance score comparing a mouse model to a reference human study with pathway-specific analysis; (2) We will extend the method to allow simultaneous resemblance analysis of multiple competing animal models and human studies, and develop downstream bioinformatics tools; (3) We will apply the methods to two on-going mouse models in breast cancer and depression, generate new hypotheses and validate the findings. Successful completion of these aims will provide practical guidance to scientists in many disease areas to best utilize animal models for understanding mechanisms and seeking novel treatments.
Model organisms and in vitro cell culture have been widely used to study disease mechanisms and to screen for therapeutic targets but debates have been continual on whether results from animal models can be representative for human studies. In this proposal, we develop a set of statistical and bioinformatic tools to identify specific pathways and regulatory subcomponents where animal models best recapitulate the human system, in order to guide best usage of these non-human models.