Discovery of drug–omics associations in type 2 diabetes with generative deep-learning models
Rosa Lundbye Allesøe, Agnete Troen Lundgaard, Ricardo Hernández Medina, Alejandro Aguayo‐Orozco, Joachim Johansen, Jakob Nybo Nissen, Caroline Brorsson, Gianluca Mazzoni, Lili Niu, Jorge Hernansanz Biel, Cristina Leal Rodríguez, Valentas Brasas, Henry Webel, Michael E. Benros, Anders Gorm Pedersen, Piotr Jaroslaw Chmura, Ulrik Plesner Jacobsen, Andrea Mari, Robert W. Koivula, Anubha Mahajan, Ana Viñuela, Juan Fernández Tajes, Sapna Sharma, Mark Haid, Mun‐Gwan Hong, Petra Musholt, Federico De Masi, Josef Korbinian Vogt, Helle Krogh Pedersen, Valborg Guðmundsdóttir, Angus G. Jones, Gwen Kennedy, Jimmy Bell, E. Louise Thomas, Gary Frost, Henrik S. Thomsen, Elizaveta Hansen, Tue H. Hansen, Henrik Vestergaard, Mirthe Muilwijk, Marieke T. Blom, Leen M. ‘t Hart, François Pattou, Violeta Raverdy, Søren Brage, Tarja Kokkola, Alison Heggie, Donna McEvoy, Miranda Mourby, Jane Kaye, Andrew T. Hattersley, Timothy J. McDonald, Martin Ridderstråle, Mark Walker, Ian Forgie, Giuseppe N. Giordano, Imre Pavo, Hartmut Ruetten, Oluf Pedersen, Torben Hansen, Emmanouil Dermitzakis, Paul W. Franks, Jochen M. Schwenk, Jerzy Adamski, Mark I. McCarthy, Ewan R. Pearson, Karina Banasik, Simon Rasmussen, Søren Brunak, Philippe Froguel, Cecilia Engel Thomas, Ragna S. Häussler, Joline W. J. Beulens, Femke Rutters, Giel Nijpels, Sabine van Oort, Lenka Groeneveld, Petra J. M. Elders, Toni Giorgino, Marianne Rodriquez, Rachel Nice, Mandy H. Perry, Susanna Bianzano, Ulrike Graefe‐Mody, Anita M. Hennige, Rolf Grempler, Patrick Baum, Hans‐Henrik Stærfeldt, Nisha Shah, Harriet Teare, Beate Ehrhardt, J. Tillner, Christiane Dings, Thorsten Lehr, Nina Scherer, Iryna Sihinevich, Louise Cabrelli, Heather Loftus, Roberto Bizzotto, Andrea Tura
Abstract
The application of multiple omics technologies in biomedical cohorts has the potential to reveal patient-level disease characteristics and individualized response to treatment. However, the scale and heterogeneous nature of multi-modal data makes integration and inference a non-trivial task. We developed a deep-learning-based framework, multi-omics variational autoencoders (MOVE), to integrate such data and applied it to a cohort of 789 people with newly diagnosed type 2 diabetes with deep multi-omics phenotyping from the DIRECT consortium. Using in silico perturbations, we identified drug-omics associations across the multi-modal datasets for the 20 most prevalent drugs given to people with type 2 diabetes with substantially higher sensitivity than univariate statistical tests. From these, we among others, identified novel associations between metformin and the gut microbiota as well as opposite molecular responses for the two statins, simvastatin and atorvastatin. We used the associations to quantify drug-drug similarities, assess the degree of polypharmacy and conclude that drug effects are distributed across the multi-omics modalities.