Researchers at InPP are developing machine learning methods for predicting phenotypic traits from genetic information of key crops. The project is led by Manisha Sirsat, from the Data Management and Risk Analysis Department, which is headed by Ricardo Ramiro, in collaboration with the Protection of Specific Crops Department, headed by Paula Oblessuc.
Over the last decade, machine learning has become part of our everyday lives, when it suggests the next song you should listen to or the restaurant you should go to. This branch of artificial intelligence is focused on building models and applications that can learn from data, in order to predict a particular outcome. For this to be possible, large amounts of data are necessary which, until recently, would preclude its application in most fields of biology. However, in the last 20 years, biology has become a data-intensive discipline, due to the revolution brought by high-throughput systems for fields as disparate as genomics and microscopy. Thus, machine learning methods are now being applied to a wide range of biological questions.
At InPP, the team is taking advantage of the availability of high-throughput genomic and phenotypic data for key phenotypes of important crops (e.g. wheat genomes and yield) and using this data to develop machine learning models that can predict the phenotype from the genotype. This approach is termed Genomic Prediction. “The aim is to develop an advanced genomic prediction tool which uses genome-wide genetic markers to predict complex traits”, states Manisha Sirsat. “This will allow us to identify genetic markers that can increase agricultural productivity and that can accelerate plant breeding programs”, adds Ricardo Ramiro.
An advanced genomic prediction tool can help accelerate plant breeding programs and increase agricultural productivity.
O artigo foi publicado originalmente em InnovPlantProtect.