Machine learned-based visualization of the diversity of grapevine genomes worldwide and in Armenia using SOMmelier

¹ Research Group of Plant Genomics, Institute of Molecular Biology of National Academy of Sciences RA, Yerevan 0014, Armenia
² Department of Genetics and Cytology, Yerevan State University, Yerevan 0025, Armenia
³ Armenian Bioinformatics Institute (ABI), Yerevan 0014, Armenia
⁴ Bioinformatics Group, Institute of Molecular Biology Institute of National Academy of Sciences RA, Yerevan 0014, Armenia, Yerevan 0014, Armenia
⁵ Interdisciplinary Centre for Bioinformatics, University of Leipzig, 04107 Leipzig, Germany

Abstract

In the proposed study three major issues have been addressed: Firstly, the diversity of grapevine accessions worldwide and particularly in Armenia, a small country located in the largely volcanic Armenian Highlands, is incredibly rich in cultivated and especially wild grapes; secondly, the information hidden in their (whole) genomes, e.g., about the domestication history of grapevine over the last 11,000 years and phenotypic traits such as cultivar utilization and a putative resistance against powdery mildew, and, thirdly machine learning methods to extract and to visualize this information in an easy to percept way. We shortly describe the Self Origanizing Maps (SOM) portrayal method called “SOMmelier” (as the vine-genome “waiter”) and illustrate its power by applying it to whole genome data of hundreds of grapevine accessions. We also give a short outlook on possible future directions of machine learning in grapevine transcriptomics and ampelogaphy.