Autism spectrum disorders (ASD) are among the most heritable developmental disorders, associated with a large number of rare genetic alterations. A critical goal of current ASD research is to deconstruct its heterogeneity into clinically homogeneous sub-set of patients, characterized by distinct neurobiological or functional deficits, amenable to precise therapeutic targeting. Fostered by the advent of resting-state fMRI (rsfMRI), human brain mapping has revealed highly heterogeneous patterns of neural synchronization (i.e. “functional connectivity”) in ASD, with evidence of inconsistent, often contrasting, patterns of over- and under-connectivity across patient cohorts. However, the origin and significance of these highly heterogeneous findings remain unclear: does genetic heterogeneity account for the observed network divergences? And can we use functional connectivity fingerprints to cluster ASD into clinically relevant sub-types? The present project leverages translationally-relevant mouse brain rsfMRI measurements to propose a first-of-its-kind decomposition of human ASD rsfMRI datasets into homogeneous subtypes, recapitulating biologically-validated “ground truth” network features identified in the mouse. To this aim, I will use a set of etiologically-relevant rsfMRI fingerprints identified in a unique mouse datasets comprising 20 ASD-associated mutations to guide clustering of a large collection of human rsfMRI datasets. Socio-cognitive profiling will be employed to probe the clinical significance and homogeneity of the identified clusters. I will next combine advanced statistical modelling and gene ontologies to explore the biological underpinnings of each identified connectivity sub-type. These investigations will lead to a novel, etiologically-relevant deconstruction of the connectional and clinical heterogeneity of ASD that may improve patient stratification, guide the identification of dysfunctional pathways and help prediction of treatment response.