Analysis of Persian Bioinformatics Research with Topic Modeling

Area of Science:

• Bioinformatics
• Information Technology
• Biological Sciences
• Natural Language Processing

Background:

• Bioinformatics is an interdisciplinary field attracting attention from IT, mathematics, and biological sciences.
• Topic modeling from natural language processing is increasingly important due to large biological datasets.

Purpose of the Study:

• To model the topic content of bioinformatics literature by Iranian researchers.
• To analyze papers indexed in the Scopus Citation Database until March 9, 2022.

Main Methods:

• Descriptive-exploratory study of 3899 papers.
• Topic modeling using Latent Dirichlet Allocation (LDA) and TF-IDF on abstracts and titles.

Main Results:

• Identified seven main topics: Molecular Modeling, Gene Expression, Biomarker, Coronavirus, Immunoinformatics, Cancer Bioinformatics, and Systems Biology.
• Systems Biology and Coronavirus represented the largest and smallest topic clusters, respectively.

Conclusions:

• The Latent Dirichlet Allocation (LDA) algorithm showed acceptable performance in classifying bioinformatics topics.
• Extracted topic clusters demonstrated excellent consistency and interconnections.