Question: Resources for machine learning prediction of phenotypes from omics data
gravatar for enricoferrero
7 days ago by
United Kingdom
enricoferrero740 wrote:


I'm looking for resources (reviews, papers, tutorials) on using machine learning approaches that discriminate between phenotypic or clinical characteristics using omics data as input.

Simplest case scenario would be to build a classifier on gene expression profiles of patients with or without a particular phenotype or clinical characteristic. Similar to a standard differential expression analysis workflow, one could then explore the most discriminating features/genes from a functional perspective.

Thank you!

ADD COMMENTlink modified 6 days ago by Kevin Blighe25k • written 7 days ago by enricoferrero740
gravatar for Kevin Blighe
6 days ago by
Kevin Blighe25k
USA / Europe / Brazil
Kevin Blighe25k wrote:

I give some related answers here:

I even go over how one can predict ethnicity using regression and PCA here: A: How to predict individual ethnicity information by using hapmap data


Whilst 'machine learning' may sound cool, the algorithms employed in the realm of machine learning do not outperform standard and well-curated regression analysis for the purposes of building classifiers.

In saying this, some of the most interesting work that I've seen in machine learning was performed in building classifiers of pathogenicity, i.e., for predicting the deleteriousness of genetic variants. For example:

  • CADD, model training using SVM with linear kernel
  • DANN, implements a deep neural network with “an input layer, a sigmoid function output layer, and three 1000-node hidden layers with hyperbolic tangent activation function
  • FATHMM-MKL, model training using SVM involving base kernels and then composite kernels


ADD COMMENTlink modified 6 days ago • written 6 days ago by Kevin Blighe25k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 582 users visited in the last hour