Proprietary tools.
Built on real research.
14 proprietary tools spanning multiomics, molecular docking, FEP simulation, enhanced sampling, AI agents, pharmacovigilance, and clinical reporting. Every package emerged from active research — validated on real data, published or in submission, containerized and version-locked.
Interactive 3D molecular visualization with binding affinity scoring, pocket detection, and interaction analysis.
bsview is Boring Science's browser-native molecular visualization platform. It renders protein structures and protein-ligand complexes from PDB input across three rendering modes — sticks, spheres, and molecular surfaces — with Vina-like binding affinity scoring and real-time interaction detection covering hydrogen bonds, hydrophobic contacts, π-stacking, and clash detection. Built-in binding pocket analysis and 2D ligand-protein interaction maps make it a complete structural analysis environment. No installation required — runs entirely in-browser via WebGL.
Upload any protein-ligand complex and get a publication-quality 2D interaction map in seconds.
bsplmapper is Boring Science's interaction analyst — upload any protein-ligand complex as PDB, CIF, or PDBQT and receive a publication-quality 2D interaction map with a full residue interaction table. Powered by PandaMap v4 with 3D-enhanced rendering, it detects hydrogen bonds, hydrophobic contacts, π-π stacking, salt bridges, halogen bonds, metal coordination, donor-π, amide-π, and alkyl-π interactions. Designed for medicinal chemists and structural biologists who need fast, publication-ready interaction analysis without manual diagram editing.
SE(3)-equivariant GNN scoring for molecular docking. Flexible receptor, allosteric site detection, ensemble docking. Pearson R: 0.88.
bsdock is Boring Science's computational docking platform, featuring a proprietary SE(3)-equivariant Graph Neural Network scoring function. Protein-ligand complexes are encoded as heterogeneous graphs with the SE(3)-equivariant architecture guaranteeing predictions invariant to rotations and translations — a physically grounded constraint that standard empirical scoring functions ignore. bsdock ships a hybrid workflow combining traditional pose generation with GNN rescoring, plus specialist modules for flexible receptor docking, allosteric site detection, and ensemble docking. Compatible with poses from any upstream docking engine.
Patient or cohort samples to actionable clinical report — automatic gene panel selection, variant classification, and FDA drug mapping.
bsclinical is Boring Science's clinical reporting engine for patient or cohort samples with a disease context. It automatically selects the right gene panel from seven disease domains — oncology, cardiology, neurology, and four more — classifies variants against known hotspots, maps findings to FDA-approved drugs, and produces a structured HTML clinical report. Not for exploratory research data. Designed for clinical teams that need fast, structured, auditable output from WGS or WES input without manual curation overhead.
End-to-end MS proteomics pipeline — raw MS files to differential proteins, PPI network, and GO/KEGG enrichment.
bsproteomics is Boring Science's mass spectrometry proteomics pipeline covering RIME-MS, AP-MS, DDA, TMT, and DIA experiments. It runs MaxQuant quantification with DEqMS differential analysis, builds STRING PPI networks, and performs GO/KEGG pathway enrichment. Protein intensity matrices can be passed downstream to bsmultiomics for multi-omics integration. Designed for proteomics teams who need a reproducible, end-to-end pipeline from raw files to publication-ready differential protein lists — without stitching together individual tools.
ATAC/ChIP FASTQs to peaks, motifs, and differential accessibility — one assay at a time, full pipeline.
bschromatin is Boring Science's deep-dive chromatin analysis tool for ATAC-seq or ChIP-seq (one assay at a time). It runs the full alignment pipeline through peak calling with MACS3 or MACS2, DiffBind differential binding analysis, HOMER motif enrichment, and deepTools bigWig tracks — all reproducible and containerized. Start here if you have raw ATAC-seq or ChIP-seq FASTQs. Outputs can feed into bsmultiomics for multi-omics integration.
Combine two or more omics layers into shared biology — use after running individual tools.
bsmultiomics is Boring Science's cross-modal integrator. Use this after running individual tools — it accepts processed outputs from bschromatin (peak matrices), bsproteomics (LFQ intensities), and raw RNA-seq FASTQs. It runs MOFA+ factor analysis to find the shared biology driving variation across all your data layers, producing interpretable latent factors and cross-modal feature weights. Does not replace per-modality tools — it connects them. Requires at least two omics modalities.
Claude-powered AI research agent with 12 OpenClaw Medical Skills — ask anything, queries live databases.
bsagent is Boring Science's AI research agent for medical and bioinformatics queries. Powered by Claude with 12 OpenClaw Medical Skills, it provides live access to PubMed literature search, ClinicalTrials.gov, NCBI Gene and ClinVar, ChEMBL drug lookup, UniProt proteins, KEGG pathways, GWAS Catalog, Open Targets gene-disease associations, drug-target interactions, and Lipinski/QED drug-likeness calculation. Designed for researchers who need fast, evidence-backed answers from live databases without manual querying across multiple portals.
Messy CSV in — reproducible, ML-ready preprocessing pipeline out. Fully automatic.
bsdatamind is Boring Science's intelligent data profiling and pipeline generation engine. Drop in any CSV and it automatically profiles schema, infers semantic types (dates, IDs, free text, categoricals, numerics, leakage candidates, target labels), detects data quality issues (missing values, duplicates, bad units, inconsistent categories), and generates a fully reproducible preprocessing pipeline ready for scikit-learn, XGBoost, LightGBM, or deep learning. Built on eight composable components: CSVIngestor, SchemaProfiler, SemanticTypeInferer, DataQualityChecker, TransformationPlanner, PipelineGenerator, Validator, and Exporter.
Predict ICU transfer, ventilation, and in-hospital mortality 6–48 hours early. AUROC 0.758 vs NEWS2 0.568. Validated on 1.33M MIMIC-IV windows.
bssentinel is Boring Science's clinical early-warning system for patient deterioration. It combines the established NEWS2 physiological score with a temporal XGBoost model trained on 1,334,773 monitoring windows from 80,442 MIMIC-IV admissions, predicting ICU transfer, mechanical ventilation, and in-hospital mortality within a 6–48 hour horizon. AUROC 0.758 (95% CI 0.755–0.761) versus NEWS2 0.568 — a DeLong z-statistic of 80.3 (p < 0.0001). Missingness indicators for high-acuity labs such as lactate are encoded as informative features, not discarded. Every prediction ships with SHAP feature attribution so clinicians understand which vitals and labs drove the alert. Manuscript in preparation for Science Advances.
Real-time intraoperative monitoring with live SIRS/qSOFA sepsis risk scoring, Random Forest deterioration alerts, and weight-based drug dosing — multi-case OR dashboard.
bsorbit is Boring Science's OR-BIT intraoperative intelligence platform. It streams live physiological data via WebSocket into a multi-case operating room dashboard, continuously scoring SIRS criteria and qSOFA sepsis risk, and running a Random Forest model for real-time deterioration detection. Integrated weight-based drug dosing calculators cover anaesthetic and critical care agents. Designed for anaesthetists and intensivists who need a single situational-awareness display across concurrent surgical cases without switching between separate monitoring systems.
From expression matrix to inferred GRN, Boolean attractors, and in-silico perturbation — gene circuit analysis end-to-end.
bscircuit is Boring Science's gene regulatory network inference and simulation platform. It runs GRNBoost2 to infer a weighted regulatory network from any gene expression matrix (CSV or TSV), identifies Boolean network attractors representing stable cell states, performs in-silico transcription factor perturbations to predict downstream gene expression changes, and exports the final network in Cytoscape-compatible SIF format for publication-quality visualization. Built for researchers studying transcriptional circuits, cell fate decisions, and disease-associated regulatory rewiring.
Visual drag-and-drop omics pipeline designer — 80+ analysis nodes covering RNA-seq, scRNA-seq, ATAC-seq, ChIP-seq, WGS/WES, and 16S microbiome.
bsprofiling is Boring Science's node-based visual pipeline canvas for designing, configuring, and exporting multi-omics analysis workflows. Built on XYFlow with 80+ preconfigured analysis nodes, it covers the full breadth of modern genomics: bulk RNA-seq (DESeq2, edgeR), single-cell RNA-seq (Seurat, Scanpy), ATAC-seq chromatin accessibility, ChIP-seq peak calling, whole-genome and whole-exome sequencing, and 16S microbiome profiling. Specialized node groups include Clinical Gene Panels (bsONCO, bsNEURO, bsCARDIO and more with disease-specific sub-panels), Proteomics, and Chromatin. Pipelines export as nf-core compatible Nextflow DSL2. No bioinformatics expertise required to assemble a reproducible pipeline.
Multi-evidence target scoring across OpenTargets, FAERS, SIDER, ChEMBL, and gene regulatory networks — tiered candidate list with PDF report.
bstarget is Boring Science's drug target and candidate prioritisation engine for early-stage BioPharma programmes. Given a gene or protein of interest, it pulls evidence from five independent sources — Open Targets association scores, FDA FAERS adverse event signals, SIDER side-effect profiles, ChEMBL compound activity data, and mechanistic gene regulatory network scores from bscircuit — and combines them into a transparent tier-ranked candidate list. Output is a structured PDF report ready for target validation or portfolio review meetings. Designed for discovery scientists who need an evidence-weighted shortlist without manual database trawling.
Search PubChem and ChEMBL, screen 14 ADMET rules, flag PAINS, and export 3D SDF libraries — interactive compound intelligence in-browser.
bschem is Boring Science's chemical intelligence platform for compound profiling and ADMET screening. Search PubChem and ChEMBL by name or SMILES, then instantly apply 14 ADMET rules covering Lipinski Ro5, Veber, REOS, QED, metabolic stability flags, and PAINS/Brenk structural alerts. RDKit handles all chemistry: 2D depiction, physicochemical property calculation (MW, logP, TPSA, HBD/HBA, rotatable bonds), and 3D SDF conformer embedding for export. The interactive compound table lets you filter, rank, and export your curated library directly into bsdock for docking or bsflow for downstream screening.
Visual node-based drug discovery workflow — connect protein fetch, binding site detection, ADMET screening, docking, and pose ranking in a single canvas.
bsflow is Boring Science's drug discovery workflow canvas, built on XYFlow for visual, drag-and-drop pipeline construction. It connects the full small-molecule discovery chain as configurable nodes: fetch a protein from RCSB PDB, detect and annotate binding sites, screen compound libraries through ADMET filters, run molecular docking via bsdock, and rank output poses by score. Workflows are saved as JSON and can be shared or re-executed reproducibly. Designed for medicinal chemists and structural biologists who want to explore a target without writing a single line of code.
ICH E2A-compliant adverse event expectedness assessment — FAERS signal lookup, FDA label review, MedDRA matching, and ICSR E2B(R3) export.
bspv is Boring Science's pharmacovigilance intelligence platform for adverse event expectedness assessment under ICH E2A. Given a drug name or RxNorm CUI and an adverse event description, it queries FDA FAERS for historical signal data, reviews the current FDA prescribing label, performs fuzzy MedDRA term matching to standardise the event, and renders a structured expectedness determination. Output exports as an ICSR-compatible E2B(R3) JSON suitable for direct regulatory submission workflows. Designed for pharmacovigilance scientists, medical safety officers, and CRO teams who need fast, auditable AE assessments without manual label review.
Alchemical FEP and thermodynamic integration for rigorous binding free energy prediction at lead optimization scale.
bsfep is Boring Science's free energy perturbation platform. It automates relative binding free energy (RBFE) and absolute binding free energy (ABFE) calculations using alchemical perturbation theory, with support for GROMACS and OpenMM backends. The workflow handles lambda window setup, Bennett Acceptance Ratio (BAR) and MBAR analysis, cycle closure correction, and uncertainty quantification. Designed for lead series where rank-ordering by docking score is insufficient — bsfep delivers thermodynamically rigorous ΔΔG predictions targeting sub-kcal/mol accuracy.
Interested in running our pipelines on your data? We partner with research groups, pharma, and diagnostics labs for pilot engagements.
"Every package ships with a test suite, a lockfile, and a container. If it works on our machine, it works on yours — and on the FDA's."