Automated Training for Algorithms That Learn from Genomic Data

Abstract EN

Supervised machine learning algorithms are used by life scientists for a variety of objectives.

Expert-curated public gene and protein databases are major resources for gathering data totrain these algorithms.

While these data resources are continuously updated, generally, theseupdates are not incorporated into published machine learning algorithms which thereby canbecome outdated soon after their introduction.

In this paper, we propose a new model ofoperation for supervised machine learning algorithms that learn from genomic data.

By definingthese algorithms in a pipeline in which the training data gathering procedure and the learningprocess are automated, one can create a system that generates a classifier or predictor usinginformation available from public resources.

The proposed model is explained using three casestudies on SignalP, MemLoci, and ApicoAP in which existing machine learning models areutilized in pipelines.

Given that the vast majority of the procedures described for gatheringtraining data can easily be automated, it is possible to transform valuable machine learningalgorithms into self-evolving learners that benefit from the ever-changing data available forgene products and to develop new machine learning algorithms that are similarly capable.