Code & Toolkits

AO-EB_batch_quant, CLI-driven Python tool for high-throughput analysis of acridine-orange/ethidium-bromide live-dead assays. It auto-discovers paired green (AO) and red (EB) TIFFs, performs dual-channel watershed segmentation, and classifies every nucleus as Live, Early Apoptotic, Late Apoptotic, or Necrotic using adaptable, self-tuning gates with a built-in red-signal safety cap. Vectorised NumPy statistics deliver 4× speed and lean memory usage, while optional overlays and 16-bit masks provide instant QC. Outputs include per-field summaries, per-cell CSVs, and a provenance JSON—ideal for screening pipelines or notebook workflows.

Vina-Batch-Docking-Windows, Windows-native wrapper that lets you screen thousands of PDBQT ligands with AutoDock Vina in a single command. The repo bundles the latest Vina binaries, a streamlined Perl script (Vina_windows.pl) for queueing jobs across all CPU cores, and starter configs so you can move from receptor.

Neubauer-Counter is a lightweight, head-less Python tool that converts any cropped Neubauer or Improved-Neubauer frame into an accurate cell tally with a single command. Powered by OpenCV’s SimpleBlobDetector and just a few dozen lines of code, it needs no GUI, no manual clicks, and no proprietary software—making it ideal for automated notebooks, servers, or CI pipelines. A couple of CLI flags let you tune blob-filters, generate red-dot overlays for QC, and even output cells · mL⁻¹.

hdock-analysis-pipeline for high-throughput analysis of HDOCK protein–peptide docking results. Automatically parses Excel outputs, computes physicochemical properties (GRAVY, net charge), generates weighted composite scores, and clusters poses by RMSD. Outputs include long-form CSV, per-complex summary, and uniform, publication-quality plots (score vs normalized RMSD, interface-frequency, contact maps), with optional anchor-residue filtering via PDB inputs.

forestplotR converts Mendelian-randomization results into journal-ready forest plots with zero Illustrator cleanup. The standalone Rscript reads a CSV of β estimates and CIs, filters Egger rows, automatically facets by outcome, and exports scalable SVGs sized precisely to single- or double-column widths. Embedded fonts via showtext guarantee true vector output, while Dark2 palette, dashed zero line, and uniform row spacing deliver polished, publication-grade aesthetics. 

compile_hdock_excel is a slim Python utility that turns raw HDOCK job URLs into a single, analytics-ready Excel workbook. It auto-detects ranked_poses.txt or the HTML table, converts the top-10 matrix to five concise rows, stacks every complex on one sheet, and appends a live hyperlink to each all_results.tar.gz archive. No GUI, no config—just list URLs, run the script, and start filtering scores, RMSDs, or interface models in seconds. Cross-platform, pure-Python, installs easily in seconds via pip.

hdock_batch Async Python CLI that uses Playwright-controlled headless Chromium to batch-submit receptor-ligand or protein-protein jobs to the hybrid template-based / template-free HDOCK server. It parses a user-supplied CSV with pandas, launches multiple browsers in parallel, and writes a timestamped run-log.csv for fully reproducible, high-throughput docking campaigns.

hmdb-endo-flagger streams HMDB’s XML dump via xml.etree.iterparse, extracting text from origin, description, biospecimens and ontology tags, then computes a raw score using curated positive/negative keyword weights. A logistic sigmoid maps that score to a 0–1 confidence, and any metabolite above a user-set threshold is flagged as human-endogenous in the TSV output.

hmdb_endogenous_animal – A fault-tolerant Python crawler that streams HMDB’s 1-GB XML dump instead of loading it, parses MetaboCards on the fly, and flags every compound classified “Endogenous → Animal.” Thread-pooled workers with automatic retry keep memory in the low-MB range and let you resume after a crash, delivering a clean two-column TSV ready for R or pandas analytics

ligand_pdb2pdbqt – One-click Windows batch pipeline for virtual-screening prep. It batch-protonates hundreds of ligand PDBs at physiological pH 7.4 using Open Babel 3, then hands each structure to AutoDockTools’ prepare_ligand4.py to generate reproducible PDBQT files. Dependencies are self-checked, scratch files are cleaned automatically, and naming conventions stay consistent—so you can move straight to Vina without manual tinkering.

receptor-peptide-interface-mapper – A high-throughput Biopython CLI that scans any mix of PDB or mmCIF files (even .gz) and pinpoints all receptor residues within a user-defined Ångström cutoff of peptide chains. Multithreaded processing lets it crunch hundreds of complexes in minutes; the six-column CSV output drops seamlessly into downstream stats or visualization pipelines. Pure Python 3, no heavy dependencies.