Fixing release for zenodo

v0.1.1

Release v0.1.1

PIGLasso v0.1.0

PIGLasso v0.1.0 — Initial Release

Prior-Informed Graphical Lasso pipeline for burn injury transcriptomics, built on top of NODIS.

PIGLasso does not perform statistical inference. Edge-level p-values, FDR control, and confidence intervals are provided by NODIS, which sits above PIGLasso in the stack. This repository is the domain-specific pipeline that applies PIGLassoEstimator (from NODIS) to burn injury transcriptomics data and extends it with biological prior construction, network diffusion, node knockouts, pathway annotation, and GSEA.

---

Features

• Stability-selection GGM with per-edge biological prior weighting: λ_ij = λ · (1 − α · P_ij)
• Prior construction pipeline — STRING PPI + KEGG pathway co-membership (steps 1 → 2a/2b/2c → 3 → 4)
• Network diffusion (heat-kernel) for transcriptional hub identification
• Node knockouts with knockout–pathway cross-reference module
• Pathway annotation & GSEA on stability scores and acute-phase delta vectors (T0+Early+Mid, 0–84 h)
• Burn-specific preprocessing for GSE182616 (burn injury) and GSE236713 (healthy + SIRS) on matched Agilent GPL17077 arrays
• CLI: piglasso run / prior / diffuse / knockout
• HPC-ready: SLURM jobs for Snellius (SURF) with full prior-build submission chain

---

Installation

pip install nodis        # required — provides PIGLassoEstimator
pip install piglasso==0.1.0

For R-based pathway modules (GSEA, gene annotation):

pip install "piglasso[r]==0.1.0"
# In R: install.packages(c("clusterProfiler", "org.Hs.eg.db", "fgsea"))

---

Datasets

Accession	Platform	Role
GSE182616	Agilent GPL17077	Primary burn injury analysis
GSE236713	Agilent GPL17077	Healthy + SIRS baseline, diffusion reference

Data are not included in the repository. See DATA.md for download instructions.

---

Authors

Roland V. Bumbuc, Zoe Azra Blei, Marcello Baryllii — Amsterdam UMC, University of Amsterdam