CorrMapper is an online research tool for the integration and visualisation of complex biomedical and omics datasets.

It allows users to:
  • map clinical metadata onto the omics datasets using an automatically generated dashboard interface,
  • perform feature selection on the omics datasets using one of the clinical metadata variables,
  • infer robust correlations between the selected features of one or two omics datasets,
  • visualise and analyse the networks of these correlations using highly interactive modules.
CorrMapper is a data exploration and hypothesis generation tool. It does not try to automate statistical inference, or provide predicitve models. It is simply making the simultaneous exploration of omics datasets and clinical metadata easier by reducing the number of predictors to the clinically relevant ones, and by providing novel and interactive visualisation modules.

Very importantly, if you would like to use the features that were selected by CorrMapper for modelling, you must ensure that the model is built and validated on new data, that was not included in the feature selection. Otherwise the generalisation error of your model will be underestimated. See Chapter 7.10.2 The Wrong and Right Way to Do Cross-validation in The Elements of Statistical Learning on page 245.

The demos and the intro video use the following breast cancer study from Chin et al to demonstrate the various components of the app:
Chin, Koei, et al. "Genomic and transcriptional aberrations linked to breast cancer pathophysiologies." Cancer cell 10.6 (2006): 529-541.

CorrMapper was developed by Daniel Homola during his PhD at Imperial College London, with the generous support of the STRATiGRAD programme and the Wellcome Trust.

Funders