Biology subjects

Lithgow-Serrano, O. W.

Publications and source records attributed to Lithgow-Serrano, O. W..

3 recordsLinked to original sources

bioRxiv2020

Lisen&Curate: A platform to facilitate knowledge tools for curation of regulation of transcription initiation in bacteria

The amount of published papers in biomedical research makes it rather impossible for a researcher to keep up to date. This is where machine processing of scientific publications could contribute to facilitate the access to knowledge. How to make use of text mining capabilities and still preserve the high quality of manual curation, is the challenge we focused on. Here we present the Lisen&Curate system designed to enable current and future NLP capabilities within a curation environment interface used in curation of literature on the regulation of transcription initiation in bacteria. The current version extracts regulatory interactions with the corresponding sentences for curators to confirm or reject accelerating their curation. It also uses an embedded metrics of sentence similarity offering the curator an alternative mechanism of navigating through semantically similar sentences within a given paper as well as across papers of a pre-defined corpus of publications pertinent to the task. We show results of the use of the system to curate literature in E. coli as well as literature in Salmonella. A major advantage of the system is to save as part of the curation work, the precise link for every curated piece of knowledge with the corresponding specific sentence(s) in the curated publication supporting it. We discuss future directions of this type of curation infrastructure.

bioinformatics↗

bioRxiv2020

L-Regulon: A novel "soft-curation" approach supported by a semantic enriched reading for RegulonDB literature.

Manual curation is a bottleneck in the processing of the vast amounts of knowledge present in the scientific literature in order to make such knowledge available in computational resources e.g., structured databases. Furthermore, the extraction of content is by necessity limited to the pre-defined concepts, features and relationships that conform to the model inherent in any knowledgebase. These pre-defined elements contrast with the rich knowledge that natural language is capable of conveying. Here we present a novel experiment of what we call "soft curation" supported by an ad-hoc tuned robust natural language processing development that quantifies semantic similarity across all sentences of a given corpus of literature. This underlying machine supports novel ways to navigate and read within individual papers as well as across papers of a corpus. As a first proof-of-principle experiment, we applied this approach to more than 100 collections of papers, selected from RegulonDB, that support knowledge of the regulation transcription initiation in E. coli K-12, resulting in L-Regulon (L for "linguistic") version 1.0. Furthermore, we have initiated the mapping of RegulonDB curated promoters, promoters, to their evidence sentence in the given publication. We believe this is the first step in a novel approach for users and curators, in order to increase the accessibility of knowledge in ways yet to be discovered.

bioinformatics↗

bioRxiv2017

Similarity corpus on microbial transcriptional regulation

The ability to express the same meaning in different ways is a well known property of natural language. This amazing property is the source of major difficulties in natural language processing. Given the constant increase in published literature, its curation and information extraction would strongly benefit by efficient automatic processes, for which, corpora of sentences evaluated by experts is a valuable resource. Given our interest in applying such approaches to the benefit of curation of the biomedical literature, specifically about gene regulation in microbial organisms, we decided to build a corpus with graded textual similarity evaluated by curators, and designed specifically oriented to our purposes. Based on the predefined statistical power of future analyses, we defined features of the design including sampling, selection criteria, balance, and size among others. A non-fully crossed-design was performed for each pair of sentences by 3 evaluators from 7 different groups, adapting the SEMEVAL scale to our goals in four successive iterative sessions with a clear improvement in the consensuated guidelines and inter-rater-reliability results. Alternatives for the corpus evaluation are widely discussed. To the best of our knowledge this is the first similarity corpus in this domain of knowledge. We have initiated its incorporation in our research towards high throughput curation strategies based in natural language processing.

bioinformatics↗