C-STDP: Causal Spike-Timing Dependent Plasticity for GRN Inference

C-STDP is a computational biology pipeline designed to infer directed, causal Gene Regulatory Networks (GRNs) from transcriptomic time-series data. It adapts the neurobiological principle of Spike-Timing Dependent Plasticity (STDP) to gene expression, treating transcriptional bursts as "spikes" to identify temporal precedence patterns.

Symposium Pivot: Mapping the COVID-19 Host Response

The project has been specialized for Infectious Disease Analysis, specifically mapping the temporal cascade of the host immune response to SARS-CoV-2. By focusing on the timing of activation events, C-STDP identifies upstream regulators that drive inflammatory and antiviral pathways.

Key Features

• Sudden-Onset Detection: Implements derivative-based Z-score encoding to identify transcriptional "bursts" (rate of change) rather than absolute amplitude.
• Immune-Aware Preprocessing: Automated pipeline for gene ID harmonization (Ensembl -> Symbol) and filtering via a curated "Immune Gene Universe" (KEGG, Reactome, GO).
• Vectorized C-STDP: High-performance inference engine capable of processing thousands of gene pairs in minutes.
• Temporal Cascade Visualization: Generates infection timelines showing the chronological order of host factor activation and their causal dependencies.

Repository Structure

C-STDP/
├── data/                   # Raw and processed transcriptomic data (GSE215865, GSE157859)
├── src/                    # Core Library Package
│   ├── core.py             # Main CausalSTDP algorithm
│   ├── utils/              # Preprocessing, mapping, and pathway annotation
│   └── visualization/      # Temporal cascade plotting
├── scripts/                # Functional Scripts
│   ├── inference/          # Spike encoding and vectorized GRN inference
│   └── visualization/      # Cascade generation
├── results/                # Output Artifacts
│   └── visuals/            # Generated cascade plots and heatmaps
├── docs/                   # Documentation
    ├── algorithm.md        # Detailed description of the infectious disease framing
    └── pipeline/           # Historical algorithmic documentation

Methodology

The C-STDP algorithm treats temporal precedence as a proxy for causality. For any pair of genes (A, B), if Gene A consistently activates before Gene B, the link A $\to$ B is strengthened. This allows for the reconstruction of the host factor activation timeline.

Detailed algorithmic documentation is available in docs/algorithm.md.

Usage

1. Preprocess Immune Subset

python3 src/utils/preprocess.py

2. Infer GRN (Vectorized)

python3 scripts/inference/run_real_data_cstdp.py

3. Generate Visualizations

python3 scripts/visualization/generate_cascade.py

Results

The pipeline identifies top regulators associated with:

• Viral mRNA Synthesis (e.g., POLR2C, NUP153)
• Interferon Induction (e.g., NLRC5, IFI35)
• Stress Response (e.g., HMOX1)

Author

MD. Arshad Email: arshad10867c@gmail.com Affiliation: Jamia Millia Islamia

License

MIT License. See LICENSE for details.