Rare cell population identification method and system based on heterograph neural network

Abstract

The invention discloses a rare cell population identification method and system based on a heterograph neural network, and the method comprises the steps: carrying out the standardization processing of cell sequencing data, and obtaining a standardized gene expression matrix; performing graph construction processing on the standardized gene expression matrix to obtain a cell gene heterograph; performing node feature coding processing on the cell gene heterograph according to a heterograph neural network to obtain cell node features; and carrying out rare cell population classification identification processing on the cell node features to obtain a rare cell population identification result. The embodiments of the present application can accurately capture rare cell features, significantly improve recognition performance, and can be widely applied to the technical field of computers.

Images