Comparative Analysis of Preimplantation Gene Regulation in Homo Sapiens and Mus Musculus Using Multi-Layer Perceptron-Based Neural Networks

This study investigated gene expression changes during preimplantation embryo development in Homo sapiens and Mus musculus. RNA-sequencing data from 31 human and 42 mouse samples across various developmental stages (oocyte, zygote, 2-cell, 4-cell, 8-cell, and blastocyst) were analyzed to identify differentially expressed genes (DEGs). Intraspecies comparisons revealed no shared DEGs between humans and mice. A neural network model based on a multi-layer perceptron (MLP) was employed to prioritize the top 100 genes with significant expression changes. These genes were then mapped to orthologous counterparts between species, focusing on those with over 80% sequence identity and conserved gene order. Five essential genes were chosen for every stage of preimplantation the model yielded ROC-AUCs of 0.99 and 1.00, respectively, with average cross-validation scores of 0.83 for humans and 0.81 for mice. The model yielded ROC-AUCs of 0.99 and 1.00, respectively, with average cross-validation scores of 0.83 for humans and 0.81 for mice. This comparative study gives insights into conserved gene expression mechanisms governing early mammalian development across species, increasing our knowledge of the invaluable changes in preimplantation stages.