Abstract
The rapid expansion of COVID-19 literature presents a significant challenge for researchers attempting to track gene-disease associations. This study employs "intelligent bibliometrics"—a combination of natural language processing, network analytics, and semantic drift analysis—to map the evolution of over 5,600 human-focused genetic research papers. By identifying key institutional players and emerging research topics, the study highlights critical genes like FURIN and the OAS family as high-priority targets for future therapeutic intervention.
Introduction
The COVID-19 pandemic triggered an unprecedented surge in scientific output, creating a "productivity downside" where manual tracking of the shifting genetic landscape became nearly impossible. Traditional bibliometrics, which typically rely on citation counts, fail to capture the semantic nuances and thematic evolution of a field. This research addresses this gap by developing a framework to identify key institutional contributors, dominant research topics, and, crucially, the specific genes emerging as central actors in the COVID-19 response.
Methodology: Intelligent Bibliometrics
The study utilizes a three-stage "intelligent bibliometrics" approach to move beyond simple publication counts.
- Co-occurrence Analysis: Using VOSviewer, the authors mapped collaborations between institutions and the relationships between scientific terms to visualize the global research network.
- Scientific Evolutionary Pathways (SEP): This novel method employs K-means clustering and "semantic drift" analysis. By treating documents as streaming vectors of terms, the authors could observe how initial research topics evolved into more specialized descendant topics over time.
- Bio-entity Network Analysis: Utilizing Pubtator, a deep-learning tool, the team extracted specific biological entities (genes, chemicals, diseases) from the text. Graph theory metrics—including Degree, Closeness, and Betweenness Centrality—were combined with a custom Intersection Ratio (IR) to rank genes based on their centrality and specificity to COVID-19.
Data Sources and Validation
The primary dataset consisted of 5,632 human-focused COVID-19 genetic publications extracted from PubMed up to March 2021. To ensure comprehensiveness, the authors compared PubMed results with the Web of Science (WoS) database, finding a 93% overlap. The study also integrated reference dictionaries such as MeSH (Medical Subject Headings), the NCBI gene dictionary for Homo sapiens, and the dbSNP database for genetic variants.
Key Insights and Trends
Global Collaboration Patterns
The analysis revealed a distinct split in collaboration strategies. While the United States and the United Kingdom established broad, central international networks, Chinese institutions exhibited high levels of domestic collaboration with relatively sparse international links during the study period. This trend may reflect a combination of policy influences and pandemic-related border restrictions.
Evolution of Research Topics
Research themes did not merely grow; they branched significantly. Initial studies focused on comparative analysis (SARS-CoV-2 vs. SARS/MERS) before diverging into two primary fronts: viral molecular structure/genetics and the human host's immune response and vaccine development.
Identification of Emerging Genes
While ACE2 remains the most frequently discussed gene, the "intelligent" approach identified a cluster of emerging genes—OAS1, OAS2, OAS3—that are highly specific to COVID-19. These genes, involved in the interferon response, have been linked to illness severity, marking them as critical targets for drug discovery.
Results
The United States (1,811 papers) and China (1,030 papers) emerged as the leading contributors to the field, with the University of California and the University of Texas being the most prolific institutions. The research landscape is partitioned into four main communities: virus infection mechanisms, testing, gene expression/immune response, and clinical manifestation.
The study observed a "blooming" trend in publications until August 2020, followed by a slight downward trend as the field shifted from emergency response to more consolidated, long-term genetic studies. Furthermore, the study identified "Group 1" genes (e.g., IFITM2, SAT1, S100A8) as highly specific emerging entities appearing in high-impact recent research.
Limitations and Future Directions
The study's reliance on PubMed, while comprehensive for biomedicine, may exclude interdisciplinary work or pre-prints. Additionally, the authors emphasize that their data-driven "Importance vs. Specificity" map requires clinical validation and "wet lab" testing by medical experts to confirm the therapeutic potential of the identified genes. Future research aims to integrate more curated medical knowledge bases to further refine the bibliometric analysis.
Summary updated