Fig. 1.

Scheme of the work-flow used to obtain the new catalogue of putative synaptic genes. First, we trained two models with the original training set, one with the original training scheme and one with the new training scheme (Fig. 2). Then we compared the enrichment in new synaptic genes (NSG, see Methods for definition of enrichment) of the catalogues resulting from each method. After testing that the new training scheme improved the prediction (Fig. 3), we incorporated the 79 NSG to the training set, trained a new model with the new training scheme and obtained a new catalogue of putative synaptic genes