Output Format

GeneInsight produces a comprehensive set of output files containing the results of gene set analysis. This section describes the structure and content of these outputs.

Directory Structure

The output directory contains:

output/
├── index.html # Interactive HTML report
├── enrichment.csv # Gene enrichment data from StringDB
├── documents.csv # Document descriptions for topic modeling
├── topics.csv # Topic modeling results
├── prompts.csv # Generated prompts for API
├── api_results.csv # Results from API calls
├── summary.csv # Summary of topic modeling and enrichment
├── enriched.csv # Hypergeometric enrichment results
├── filtered.csv # Final filtered topics
├── metadata.csv # Run information and parameters
└── geneinsight_results.zip # All results in compressed format

File Descriptions

enrichment.csv

Contains gene enrichment data retrieved from the STRING database:

• Gene identifiers

• Associated terms and descriptions

• Statistical significance measures

documents.csv

The corpus of gene-specific descriptions used for topic modeling:

• Document IDs

• Gene associations

• Full text of annotations

topics.csv

Results from the topic modeling phase:

• Topic IDs

• Representative terms

• Frequency and distribution information

prompts.csv

Generated prompts sent to the language model API:

• Prompt IDs

• Full text of prompts

• Associated topics

api_results.csv

Responses received from the language model API:

• Formatted biological theme descriptions

• Metadata about API calls

• Processing timestamps

summary.csv

Consolidated summary of topic modeling and enrichment results:

• Theme IDs

• Representative genes

• Statistical metrics

enriched.csv

Results of hypergeometric enrichment analysis:

• Theme identifiers

• p-values and false discovery rates

• Enrichment scores

filtered.csv

Final filtered topics based on statistical significance:

• Selected theme IDs

• Ranking information

• Selection criteria

metadata.csv

Information about the analysis run:

• Tool version

• Run parameters

• Execution timestamps

Interactive HTML Report

The index.html file provides a entry point to the html report which provides an interactive visualization of the analysis results.

Theme Pages

Dedicated pages for each identified theme featuring:

• Theme descriptions generated by the language model

• Associated genes with links to reference databases

• Statistical significance metrics

Gene Set Visualizations

Heatmaps and network diagrams showing:

• Gene presence across references

• Relationships between genes and biological themes

• Statistical enrichment patterns

Download Interface

Interactive interface to download:

• Specific themes of interest

• Associated gene sets

• Customized subsets of results

Exploring the Results

After running GeneInsight, navigate to the output directory to find:

• HTML Report: Open index.html to view the interactive visualization

• CSV Files: Explore detailed results in the various output files

• ZIP Archive: A compressed version of all outputs for easy sharing

Understanding the Output

The HTML report includes:

1. Topic Map: A visual representation of identified biological themes

The topic map allows you to intuitively see larger groups of related biological themes without having to manually cross-reference multiple ontologies. Themes that appear close together in the visualization share biological meaning, functional relationships, or relevance to similar biological processes, even if they come from different ontology sources.

This visualization is particularly valuable for identifying unexpected relationships between biological themes that might not be apparent when examining individual ontology terms in isolation.

2. Gene Set Visualizations: Heatmaps showing gene presence across references

These heatmaps are particularly useful for:

1. Identifying which genes contribute most strongly to specific biological themes

2. Discovering how terms from commonly used ontologies (GO, HPO) relate to each other through shared gene associations

3. Finding unexpected gene-theme associations that might suggest novel biological functions

4. Prioritizing genes for experimental validation based on their prominence across multiple themes

The hierarchical clustering of both genes and themes helps reveal patterns that might not be obvious when examining individual gene-theme pairs, providing a systems-level view of the biological relationships in your data.

3. Theme Pages and Gene-Level Information: Detailed exploration of each identified theme

These detailed theme pages serve as comprehensive reference sheets for each biological theme identified in your analysis. They bring together information that would otherwise require searching across multiple databases and resources.

The theme pages consolidate:

1. Detailed enrichment information for each biological theme

2. STRING-DB annotations for associated genes

3. Ontology annotations across different systems (GO, HPO)

Additionally, the integration with NCBI’s API provides gene descriptions directly within the report. This consolidation of information allows researchers to quickly assess the biological relevance of each theme and its associated genes without having to manually search across multiple resources.

Practical Applications

This integrated report is designed to support several common research workflows:

1. Hypothesis Generation: Identify unexpected relationships between genes and biological processes that may suggest novel mechanisms

2. Candidate Prioritization: Rank genes for experimental validation based on their prominence across multiple biological themes

3. Pathway Analysis: Understand how genes of interest relate to established biological pathways and functions

4. Cross-Ontology Interpretation: Bridge the gap between different ontology frameworks (GO, HPO) through their shared gene associations

5. Data Integration: Combine your experimental findings with established knowledge from public databases