ZH Seminars Bioinformatics - Damian Szklarczyk

Analysis of date-tagged protein-protein associations derived from scientific literature.

Abstract Co-occurence based text-mining of protein-protein association in scientific literature proved to be the largest, source of high and low confidence PPI eclipsing well established data-sources such as manually curated databases or direct experimental observations by the factor of 2 an 3 respectively. Invariably, but also understandably, no method incorporates time and source of the text-mined associations as a potential predictive variable, as it may lead to the introduction of additional biases. In this work we have tagged each of the millions text derived PPIs with the exact date and source of their publication and explored what potential insights can be gained from such stratified data.

Damian  2019-05-22
Edit on Github...