A novel in silico transcription factor perturbation simulator uncovers diverse genetic architecture of brain disorders

Brain disorders with complex genetic architectures pose challenges in studying their mechanisms and developing treatments. As experimental approaches fall short due to these complexities, we developed TFdisc, a novel in silico transcription factor (TF) perturbation simulator that uses wild-type single-cell RNA sequencing (scRNA-seq) data to emulate corresponding scRNA-seq transcriptomes post-TF perturbations. Validation confirms TFdisc’s accuracy in reconstructing gene regulatory networks, identifying differentially expressed genes, and predicting alterations in cell identity and lineage differentiation post-TF perturbations. Applying TFdisc to simulate perturbations of individual risk TFs associated with brain disorders reveals distinct molecular and cellular mechanisms. Further simulation of multiple TF simultaneous perturbations uncovered previously uncharacterized genetic architectures for SCZ and ASD, proposing a “jigsaw mechanism” for SCZ and a “monolithic mechanism” for ASD. Further analysis of ASD risk TFs identified clusters whose perturbation leads to differential expression of genes related to synaptic function, cognition, learning, and histone modifications, with varying impacts correlated with autism severity.