[Paper Review] Filtering Microarray Correlations by Statistical Literature Analysis Yields Potential Hypotheses for Lactation Research

Maurice HT Ling, Christophe Lefèvre+1 more|Jan 2, 2009

Biomedical Text Mining and Ontologies39 references3 citations

TL;DR

This study introduces a statistical literature analysis method that filters microarray gene correlations by identifying significant protein co-occurrence patterns in scientific literature, outperforming traditional Poisson-based methods. By integrating microarray data with literature-derived co-occurrence patterns, the approach identifies 7 novel, potentially functional protein-protein interactions relevant to lactation biology not previously reported in the literature.

ABSTRACT

Our results demonstrated that a previously reported protein name co-occurrence method (5-mention PubGene) which was not based on a hypothesis testing framework, it is generally statistically more significant than the 99th percentile of Poisson distribution-based method of calculating co-occurrence. It agrees with previous methods using natural language processing to extract protein-protein interaction from text as more than 96% of the interactions found by natural language processing methods to overlap with the results from 5-mention PubGene method. However, less than 2% of the gene co-expressions analyzed by microarray were found from direct co-occurrence or interaction information extraction from the literature. At the same time, combining microarray and literature analyses, we derive a novel set of 7 potential functional protein-protein interactions that had not been previously described in the literature.

Motivation & Objective

• To identify biologically relevant protein-protein interactions in lactation research by analyzing co-occurrence patterns in scientific literature.
• To improve upon existing literature-based interaction detection methods by using a statistically robust approach rather than relying on hypothesis testing frameworks.
• To bridge the gap between microarray-derived gene expression correlations and literature-derived protein interactions by filtering and validating co-occurrence signals.
• To generate novel, testable biological hypotheses for lactation-related pathways using combined microarray and literature analysis.
• To evaluate the effectiveness of the 5-mention PubGene method against established natural language processing techniques and statistical models.

Proposed method

• Applies the 5-mention PubGene method, which identifies protein co-occurrence in abstracts based on frequency thresholds, to filter statistically significant gene interactions.
• Compares the 5-mention PubGene method against a Poisson distribution-based co-occurrence model to assess statistical significance.
• Validates the method by comparing its results with those from natural language processing (NLP) tools that extract protein-protein interactions from text.
• Integrates microarray-derived gene co-expression data with literature-derived protein co-occurrence patterns to identify potential functional interactions.
• Filters and prioritizes candidate interactions based on statistical significance and overlap with known biological pathways.
• Uses a threshold of at least five co-occurrences of protein pairs in abstracts to define significant associations, enhancing reliability over lower-frequency signals.

Experimental results

Research questions

• RQ1 How does the 5-mention PubGene method compare in statistical significance to Poisson-based co-occurrence models in detecting protein-protein interactions?
• RQ2 To what extent do the protein co-occurrence patterns detected by the 5-mention method overlap with those identified by NLP-based interaction extraction tools?
• RQ3 What proportion of microarray-derived gene co-expression pairs can be explained by direct literature co-occurrence or known protein interactions?
• RQ4 Can integrating microarray data with literature co-occurrence patterns reveal novel, biologically plausible protein-protein interactions not previously described?
• RQ5 Are the novel interactions identified through this method potentially relevant to lactation biology based on functional context?

Key findings

• The 5-mention PubGene method demonstrated statistical significance exceeding the 99th percentile of the Poisson distribution-based co-occurrence method.
• Over 96% of protein-protein interactions identified by NLP tools were also detected by the 5-mention PubGene method, indicating strong agreement between the two approaches.
• Less than 2% of microarray-derived gene co-expression pairs were supported by direct co-occurrence or interaction information extracted from the literature.
• The integration of microarray and literature analysis yielded 7 novel protein-protein interactions not previously described in the literature.
• These 7 novel interactions were functionally plausible and potentially relevant to lactation biology, suggesting the method's utility in generating testable hypotheses.
• The study confirms that literature co-occurrence patterns, when filtered by statistical significance, can effectively prioritize biologically relevant interactions from high-throughput data.