Phenotypic variations vary broadly between individuals in any biological species. Understanding the genetics controlling these phenotypes still presents a difficult challenge. Beyond known contributions from genomics, many confounding components could also affect the interpretation of observed phenotypes. From a computational biology perspective, constructing a linear model is the conventional approach to link phenotypic variation with genomic data such as molecular markers. However, inaccurate measurement of phenotypes, insufficient sample sizes, limited explainable data types, and incomplete genotype-phenotype association/prediction algorithms can lead to missing out on other useful information.
Machine learning methods can automatically learn from a large scale of training data and capture signals to make accurate decisions. Many research perspectives including medical imaging, computer vision-based phenotyping, genome-wide association, high-dimensional genotype/phenotype data processing have shown their critical demands on machine learning. Exploiting data derived from diverse layers using machine learning methodologies have the potential to facilitate the investigation of the genetics underlying phenotypic changes.
This Research Topic focuses on, but is not limited to:
• Molecular signatures on phenotype prediction using machine learning algorithms;
• Novel machine learning models on associating phenotypes with multi-omics data;
• Trait discoveries using machine learning techniques to connect genetics;
• Reviews of recent machine learning applications on phenotype prediction.
We welcome Original Research and Review articles and encourage data and code to be freely available to the public. Special thanks to
Yuan Liu, from Shanghai Jiao Tong University School of Medicine, whose help was indispensable for the formation of the project.
Phenotypic variations vary broadly between individuals in any biological species. Understanding the genetics controlling these phenotypes still presents a difficult challenge. Beyond known contributions from genomics, many confounding components could also affect the interpretation of observed phenotypes. From a computational biology perspective, constructing a linear model is the conventional approach to link phenotypic variation with genomic data such as molecular markers. However, inaccurate measurement of phenotypes, insufficient sample sizes, limited explainable data types, and incomplete genotype-phenotype association/prediction algorithms can lead to missing out on other useful information.
Machine learning methods can automatically learn from a large scale of training data and capture signals to make accurate decisions. Many research perspectives including medical imaging, computer vision-based phenotyping, genome-wide association, high-dimensional genotype/phenotype data processing have shown their critical demands on machine learning. Exploiting data derived from diverse layers using machine learning methodologies have the potential to facilitate the investigation of the genetics underlying phenotypic changes.
This Research Topic focuses on, but is not limited to:
• Molecular signatures on phenotype prediction using machine learning algorithms;
• Novel machine learning models on associating phenotypes with multi-omics data;
• Trait discoveries using machine learning techniques to connect genetics;
• Reviews of recent machine learning applications on phenotype prediction.
We welcome Original Research and Review articles and encourage data and code to be freely available to the public. Special thanks to
Yuan Liu, from Shanghai Jiao Tong University School of Medicine, whose help was indispensable for the formation of the project.