Predicting Future Scientific Discoveries Based on a Networked Analysis of the Past Literature

Abstract

We present KnIT, the Knowledge Integration Toolkit, a system for accelerating scientific discovery and predicting previously unknown protein-protein interactions. Such predictions enrich biological research and are pertinent to drug discovery and the understanding of disease. Unlike a prior study, KnIT is now fully automated and demonstrably scalable. It extracts information from the scientific literature, automatically identifying direct and indirect references to protein interactions, which is knowledge that can be represented in network form. It then reasons over this network with techniques such as matrix factorization and graph diffusion to predict new, previously unknown interactions. The accuracy and scope of KnIT's knowledge extractions are validated using comparisons to structured, manually curated data sources as well as by performing retrospective studies that predict subsequent literature discoveries using literature available prior to a given date. The KnIT methodology is a step towards automated hypothesis generation from text, with potential application to other scientific domains.

IBM Watson: How Cognitive Computing Can Be Applied to Big Data Challenges in Life Sciences Research

“…PKN1 and NEK1 were 2 kinases highly ranked by Watson as having the potential to phosphorylate P53. 32 Lab experiments at Baylor College of Medicine suggested that these kinases could phosphorylate P53 in both in vitro experiments and in tests on human cells. Further experiments are being conducted to test the activity of these kinases in organisms.…”

“…In addition to visualizing expressed patterns, machine learning and predictive analytics generate hypotheses about relationships; in effect "inferring" novel connections not yet supported by overt statements in the literature as was done in a project with Baylor College of Medicine where hypotheses of new kinases that could phosphorylate TP53 were generated out of existing medical literature. 31,32 If the observation and interpretation of concepts are the foundation for discovery in a human's cognitive process, the next step is evaluation. 30 Humans have the ability to evaluate evidence and apply it to solve different types of problems.…”

“…attributes of a cognitive system could potentially aid researchers in making connections out of large datasets faster than and potentially aid them in making connections that they may not have otherwise considered. 31,32,36 To determine where cognitive systems might add the most value, Watson should be applied to a breadth of research questions. Although Watson has been applied to research on cancer kinases and drug repurposing, other projects such as predicting combinations of genes or proteins that as a group may play a role in disease onset or progression should also be attempted.…”

A systematic review on literature-based discovery workflow

“…Hence, selecting the suitable data type in the papers in crucial as they represent different perspectives (Lee et al, 2015) and information content (Kostoff et al, 2004) and mainly depends on the objective of the research. Furthermore, Nagarajan et al (2015) have discovered that the LBD performance mainly depends on the richness of the information being used. Apart from research papers, several approaches have experimented the LBD process with other traditional data types such as patents and clinical case reports.…”

“…For example, some studies (Rastegar-Mojarad et al, 2015;Cheung, Ouellette & Wasserman, 2012) have used drug-disease interactions in Comparative Toxicogenomics Database (CTD) to validate their results. Similarly, other databases such as SIDER2 (Shang et al, 2014), GEO (Faro, Giordano & Spampinato, 2011), GAD (Seki & Mostafa, 2009), StringDB (Nagarajan et al, 2015) have also been used for validation.…”