Germ cell tumors (GCT) might undergo transformation into a somatic-type malignancy (STM), resulting in a cell fate switch to tumors usually found in somatic tissues, such as rhabdomyosarcomas or adeno Show more
Germ cell tumors (GCT) might undergo transformation into a somatic-type malignancy (STM), resulting in a cell fate switch to tumors usually found in somatic tissues, such as rhabdomyosarcomas or adenocarcinomas. STM is associated with a poor prognosis, but the molecular and epigenetic mechanisms triggering STM are still enigmatic, the tissue-of-origin is under debate and biomarkers are lacking. To address these questions, we characterized a unique cohort of STM tissues on mutational, epigenetic and protein level using modern and high-throughput methods like TSO assays, 850k DNA methylation arrays and mass spectrometry. For the first time, we show that based on DNA methylation and proteome data carcinoma-related STM more closely resemble yolk-sac tumors, while sarcoma-related STM resemble teratoma. STM harbor mutations in FGF signaling factors (FGF6/23, FGFR1/4) highlighting the corresponding pathway as a therapeutic target. Furthermore, STM utilize signaling pathways, like AKT, FGF, MAPK, and WNT to mediate molecular functions coping with oxidative stress, toxin transport, DNA helicase activity, apoptosis and the cell cycle. Collectively, these data might explain the high therapy resistance of STM. Finally, we identified putative novel biomarkers secreted by STM, like EFEMP1, MIF, and DNA methylation at specific CpG dinucleotides. Show less
Chemical descriptors encode the physicochemical and structural properties of small molecules, and they are at the core of chemoinformatics. The broad release of bioactivity data has prompted enriched Show more
Chemical descriptors encode the physicochemical and structural properties of small molecules, and they are at the core of chemoinformatics. The broad release of bioactivity data has prompted enriched representations of compounds, reaching beyond chemical structures and capturing their known biological properties. Unfortunately, bioactivity descriptors are not available for most small molecules, which limits their applicability to a few thousand well characterized compounds. Here we present a collection of deep neural networks able to infer bioactivity signatures for any compound of interest, even when little or no experimental information is available for them. Our signaturizers relate to bioactivities of 25 different types (including target profiles, cellular response and clinical outcomes) and can be used as drop-in replacements for chemical descriptors in day-to-day chemoinformatics tasks. Indeed, we illustrate how inferred bioactivity signatures are useful to navigate the chemical space in a biologically relevant manner, unveiling higher-order organization in natural product collections, and to enrich mostly uncharacterized chemical libraries for activity against the drug-orphan target Snail1. Moreover, we implement a battery of signature-activity relationship (SigAR) models and show a substantial improvement in performance, with respect to chemistry-based classifiers, across a series of biophysics and physiology activity prediction benchmarks. Show less