A quantitative model to genetically interpret the histology in whole microscopy slide images is desirable to guide downstream immunohistochemistry, genomics, and precision medicine. We constructed a statistical model that predicts whether or not SPOP is mutated in prostate cancer, given only the digital whole slide after standard hematoxylin and eosin [H&E] staining. Using a TCGA cohort of 177 prostate cancer patients where 20 had mutant SPOP, we trained multiple ensembles of residual networks, accurately distinguishing SPOP mutant from SPOP non-mutant patients (test AUROC=0.74, p=0.0007 Fisher's Exact Test). We further validated our full metaensemble classifier on an independent test cohort from MSK-IMPACT of 152 patients where 19 had mutant SPOP. Mutants and non-mutants were accurately distinguished despite TCGA slides being frozen sections and MSK-IMPACT slides being formalin-fixed paraffin-embedded sections (AUROC=0.86, p=0.0038). Moreover, we scanned an additional 36 MSK-IMPACT patient having mutant SPOP, trained on this expanded MSK-IMPACT cohort (test AUROC=0.75, p=0.0002), tested on the TCGA cohort (AUROC=0.64, p=0.0306), and again accurately distinguished mutants from non-mutants using the same pipeline. Importantly, our method demonstrates tractable deep learning in this "small data" setting of 20-55 positive examples and quantifies each prediction's uncertainty with confidence intervals. To our knowledge, this is the first statistical model to predict a genetic mutation in cancer directly from the patient's digitized H&E-stained whole microscopy slide. Moreover, this is the first time quantitative features learned from patient genetics and histology have been used for content-based image retrieval, finding similar patients for a given patient where the histology appears to share the same genetic driver of disease i.e. SPOP mutation (p=0.0241 Kost's Method), and finding similar patients for a given patient that does not have have that driver mutation (p=0.0170 Kost's Method).cancer | molecular pathology | deep learning | whole slide image G enetic drivers of cancer morphology, such as E-Cadherin [CDH1] loss promoting lobular rather than ductal phenotypes in breast, are well known. TMPRSS2-ERG fusion in prostate cancer has a number of known morphological traits, including blue-tinged mucin, cribriform pattern, and macronuclei [5]. Computational pathology methods [6] typically predict clinical or genetic features as a function of histological imagery, e.g. whole slide images. Our central hypothesis is that the morphology shown in these whole slide images, having nothing more than standard hematoxylin and eosin [H&E] staining, is a function of the underlying genetic drivers. To test this hypothesis, we gathered a cohort of 499 prostate adenocarcinoma patients from The Cancer Genome Atlas [TCGA] 1 , 177 of which were suitable for analysis, with 20 of those having mutant SPOP (Figs 1, 2, and S1). We then used ensembles of deep 1 TCGA data courtesy the TCGA Research Network http://cancergenome.nih.gov/ A....