Assisting document triage for human kinome curation via machine learning

Abstract

In the era of data explosion, the increasing frequency of published articles presents unorthodox challenges to fulfill specific curation requirements for bio-literature databases. Recognizing these demands, we designed a document triage system with automatic methods that can improve efficiency to retrieve the most relevant articles in curation workflows and reduce workloads for biocurators. Since the BioCreative VI (2017), we have implemented texting mining processing in our system in hopes of providing higher effectiveness for curating articles related to human kinase proteins. We tested several machine learning methods together with state-of-the-art concept extraction tools. For features, we extracted rich co-occurrence and linguistic information to model the curation process of human kinome articles by the neXtProt database. As shown in the official evaluation on the human kinome curation task in BioCreative VI, our system can effectively retrieve 5.2 and 6.5 kinase articles with the relevant disease (DIS) and biological process (BP) information, respectively, among the top 100 returned results. Comparing to neXtA5, our system demonstrates significant improvements in prioritizing kinome-related articles as follows: our system achieves 0.458 and 0.109 for the DIS axis whereas the neXtA5’s best-reported mean average precision (MAP) and maximum precision observed are 0.41 and 0.04. Our system also outperforms the neXtA5 in retrieving BP axis with 0.195 for MAP and the neXtA5’s reported value was 0.11. These results suggest that our system may be able to assist neXtProt biocurators in practice.

Literature Triage on Genomic Variation Publications by Knowledge-enhanced Multi-channel CNN

“…tmVar detects mentions of genomic variants in literature and normalizes them into unique dbSNP RSIDs. Although tmVar may not facilitate literature triage directly, it is quite useful for positive or negative sampling in literature triage systems [2,12,19,20,21]. The READBiomed system [19] creates term lists describing interactions, mutations, and expected effects on interactions mutations, and it trains an SVM-based literature classification model using these term lists along with a range of standard bag-of-word features.…”

“…The READBiomed system [19] creates term lists describing interactions, mutations, and expected effects on interactions mutations, and it trains an SVM-based literature classification model using these term lists along with a range of standard bag-of-word features. In [20], literature triage is achieved by different classifiers (Glmnet, SVM and CNN) for the task of Human Kinome Curation. However, the work still requires a time-consuming feature engineering, including the selection of frequency, location, and linguistic features, together with manually generated keyword groups of the genetic disease field.…”

Using deep learning to identify translational research in genomic medicine beyond bench to bedside

“…More recently, deep neural network-based approaches have shown improved results in many natural language processing (NLP) tasks including text classification (11). Neural network-based approaches such as the convolutional neural network (CNN) have been applied to assist document triage of kinome curation, genomic variation and protein–protein interactions (12–14). Lee et al (13) employed the CNN to identify publications that are relevant for variant curation.…”