The invention relates to the technical field of crossing of
bioinformatics,
medical genetics and
artificial intelligence, and discloses a
gene variation
pathogenicity automatic rating method and
system based on multi-agent game and thinking chain reasoning, and the method comprises the following steps: retrieving data related to a target variation site; based on the related data; the proposal agent, the rejecter agent and the
decision maker agent carry out multi-round iteration
gene variation
pathogenicity rating; and monitoring a rating conclusion output by the
decision maker agent and the
confidence score in real time, terminating the multi-agent game step when a preset dynamic truncation condition is met, and outputting a final
genetic variation pathogenicity rating conclusion. According to the method, model illusion is eliminated through a resistance game, unstructured literatures are analyzed by utilizing a thinking chain, and accurate and transparent rating of
gene variation pathogenicity can be realized.