*2.5. Analysis of Proteomics Data*

We used Sca ffold (Version 4.8.5) to obtain the original samples report of all TAP-MS based peptide and protein identifications. Peptide identifications were retained if they satisfied a minimum of 95.0% threshold. Protein identifications were accepted if they met greater than 50.0% threshold with a minimum of two identified peptides.

To further analyze the original Sca ffold mass spectrometry data, we used APOSTL, an integrative Galaxy pipeline for a ffinity proteomics data [26]. The following global cuto ffs were applied to 7996 interactions and generated a list of 88 high confidence interactions: SaintScore cuto ff: 0.5; FoldChange cuto ff: 0; normalized spectral abundance factor (NSAF) score cuto ff: 0.0000025.

APOSTL also interrogates the CRAPome database (http://crapome.org/), which contains common contaminants in a ffinity purification–mass spectrometry data [27]. Seventeen proteins displayed a CRAPome score >90% and were candidates to be called non-specific interactors. However, two hits were plausible as they were previously implicated in microcephaly (RAB18 and NEDD1), and two hits were found to be associated with ZIKV NS proteins in a previously published independent study (AHCYL1 and GET4) [20]. Moreover, only 2 out of 17 displayed multiple interactions, suggesting that the other 15 proteins do not constitute non-specific interactors in our assay. Therefore, we decided to retain all proteins, but have indicated the high CRAPome score in Figure 2 when appropriate.

#### *2.6. Generation of the Microcephaly-Associated Protein–Protein Interaction Network (PIN)*

We generated a microcephaly-associated PIN by searching National Center for Biotechnology Information (NCBI) ENTREZ Gene using [microcephaly] AND [Homo sapiens] as a query. This exercise led to 277 genes, which were manually curated to remove those without an Online Mendelian Inheritance in Man (OMIM) designation (i.e., pseudogenes and partially characterized loci) with a final tally of 261 genes. These genes were used as input to BisoGenet [28], which adds edges between the input nodes, to generate a microcephaly-associated network with 370 interactions (Table S13). BisoGenet settings were 'input nodes only' (Methods) and checking 'protein–protein interactions' only leaving 'Protein DNA interaction' and 'microRNA silencing interaction' unchecked. Significant interaction clusters were identified using ClusterONE (Version 1.0) [29] using the following settings: 'minimum size' = 5; 'minimum density' = 0.5; 'edge weights' = unweighted. Gene ontology (GO) enrichment of clusters was done using BINGO [30] as a Cytoscape plug-in.

## *2.7. Network Generation and GO Analysis*

Network graphics were generated with Cytoscape version 3.7.1 [31]. Each NS integrated dataset was analyzed using WebGestalt to determine the enrichment of GO terms. For each bait set (all proteins

that interact with each NS bait), the number of genes in the set that was scored for a term was obtained. The number was then divided by the number of genes in the GO database for that term to obtain an enrichment ratio. Enrichment ratios were log2-transformed to depict increase and decrease changes as numerically equal, but with an opposite sign. To allow for log transformation, enrichment ratios with a 0.0 value were replaced by half of the lowest non-zero value in the complete set. Bait sets were clustered with Cluster 3.0 using the Correlation (uncentered) metric of similarity with no filtering, and the clustering method chosen was complete linkage. It was visualized using Java TreeView v 1.1.6r4 [32].
