How nanocell-lab/metabcrosstalk works
Project Type and Purpose
• This is a bioinformatics data analysis pipeline implemented in Jupyter notebooks.
• Purpose: perform proteomics-based metabolic cross-talk analyses between human and bacterial systems. It ingests experimental data, enriches functional annotations, queries external biological databases (HOMD, MetaCyc, g:Profiler, IPA), and produces visualizations and summary tables.
File Structure Analysis
• Top-level directories:
– data/ : raw and intermediate tabular datasets (TSV, CSV, XLSX)
– src/ : four Jupyter notebooks, each corresponding to a major processing step
• Data sources under data/:
– Bacteria.tsv, HOMD/*.tsv for bacterial gene mapping
– Human Proteins.xlsx and Proteomica_Humana_CO/*.csv/.xlsx for human proteomics enrichment
• Notebooks under src/:
– 01ProteomicsHuman.ipynb: loads human proteomics, preps for enrichment
– 02FunctionalEnrichmentA.ipynb: runs GOEA (g:Profiler) and IPA analyses
– 03ProteomicsBacteria.ipynb: maps bacterial proteins to genes (HOMD)
– 04MetaCyc.ipynb: queries MetaCyc for L-glutamate pathways and counts gene occurrences
• No build scripts or server components—pipeline is orchestrated by sequential notebook execution.
README Insights
• Dependencies: Python 3, Pandas, Bioinfokit, QIAGEN IPA (external tool), g:Profiler API
• The methodology describes three stages for bacteria and two for human analyses.
• External services: QIAGEN IPA (proprietary), g:Profiler (web API), HOMD database, MetaCyc database.
a. Main Components
1. Data Ingestion Layer
– Local raw data files (data/*.tsv, *.xlsx)
– External reference databases (HOMD, MetaCyc)
2. Processing & Analysis Modules
– Notebook 01: Human proteomics data preprocessing
– Notebook 02: Functional enrichment (g:Profiler + IPA)
– Notebook 03: Bacterial gene mapping via HOMD
– Notebook 04: Pathway retrieval & gene counting in MetaCyc
3. External Services
– g:Profiler REST API
– QIAGEN IPA tool (local client or API)
– HOMD and MetaCyc (flat files or web API)
4. Output/Visualization
– Figures and tables exported by notebooks (charts, CSV, XLSX)
b. Relationships and Data Flows
• Data files feed into notebooks in sequence:
01ProteomicsHuman → 02FunctionalEnrichmentA
03ProteomicsBacteria → 04MetaCyc
• 02FunctionalEnrichmentA calls g:Profiler and IPA, merges results.
• 03ProteomicsBacteria queries HOMD dataset to map proteins → genes.
• 04MetaCyc uses gene lists to query MetaCyc pathways and aggregates counts.
• Final visualizations and summary tables are generated per notebook and stored locally.
c. Architectural Patterns and Principles
• Pipeline pattern: sequential data processing stages.
• Modularity: each notebook encapsulates a logical analysis step.
• Data provenance: raw inputs, intermediate artifacts, final outputs.
• External integration: clear boundaries where external APIs/tools are invoked.
d. Technologies / Libraries
• Python 3 and Jupyter notebooks
• Pandas for data manipulation
• Bioinfokit for statistical plots
• g:Profiler API, QIAGEN IPA
• Flat-file data sources (TSV, XLSX) for HOMD and MetaCyc