BioMate is the Innovation Partner of the Autoimmune Neuromuscular & Nerve Disorders Drug Development Summit (July 21–23, 2026 · Boston, MA). On Day 1, July 22 at 9:45 AM EST, BioMate’s Chief Strategy Officer Andrew Dike takes the stage for a talk and live demonstration — walking a real myasthenia gravis case from a public dataset to a developable drug candidate in three plain-English turns.

3
plain-English turns
36,601
genes analyzed
20
samples pseudobulked
100%
QC-gated & cited

Four Bottlenecks That Cost Every Translational Program Months

Drug development in autoimmune neuromuscular diseases — myasthenia gravis, CIDP, GBS, MMN — moves fast scientifically but slowly operationally. The same four bottlenecks appear in every program:

  • Data silos. Patient cohorts, transcriptomic datasets, and published literature live in separate systems. Pulling them together for a single analysis can take days.
  • The bioinformatics queue. Every computational task waits on a small number of specialists. Questions that should take hours can take weeks.
  • The bench-to-patient gap. Molecular signals identified in bulk sequencing rarely get connected to patient stratification or clinical hypotheses without a dedicated translation step.
  • The audit burden. Regulatory and internal review demand a defensible, fully cited, reproducible trail for every result — and assembling that trail manually is slow and error-prone.

From Question to Cited Drug Candidate in Three Turns

Our live demo shows exactly how BioMate collapses these bottlenecks — starting not from a tidy, pre-processed matrix, but from a raw public dataset in whatever shape the depositor left it in. The case: GSE227835, a single-cell RNA-seq study of AChR+ myasthenia gravis versus healthy controls.

Turn 1 — Raw public data → real differential expression

One plain-English question ingests the GEO series, auto-detects the deposit format (here, dense per-sample expression tables — not the 10x matrix files most tools assume), pseudobulk-aggregates 20 samples across 36,601 genes, and runs DESeq2. The real differentially expressed genes surface — every step logged, QC-checked, and reproducible.

Turn 2 — Confirm the biology

BioMate runs gene-set enrichment on the DE genes and cross-checks the indexed literature. On this cohort the real signal is an inflammatory, TNF-response signature (e.g. GO:0071356, cellular response to tumor necrosis factor) — the actual output of the data, with citations attached, not a scripted result.

Turn 3 — Assess a candidate drug

BioMate screens iptacopan — a clinically established complement-directed MG therapeutic — through a full ADMET safety panel: clearance, hERG, CNS permeability, hepatotoxicity, with QC gates enforced at every threshold. Molecule to developability verdict, in one turn.

The hard part isn’t the analysis — it’s the raw data.

Every dataset arrives differently: bulk matrices, 10x MEX bundles, per-sample tarballs, dense expression tables. BioMate normalizes whatever the depositor deposited, then runs genomics, pathway biology, and drug screening in one session — every finding cited, every computation reproducible, every safety gate documented. The biology you see is the data’s, not the slide’s.

Want the full engineering picture?

Those three plain-English turns run ~30 discrete phases under the hood — format detection, pseudobulk aggregation, robust dispersion estimation, output inference, and more. Read the deep dive: Inside the Three Turns — The 30-Phase Pipeline From a Raw GEO Deposit to Drug Developability.

Where to Find BioMate at the Summit

Day 1 · 9:45 AM EST Talk + live demo — main stage
Day 1 · 4:05 PM Drinks Reception
All Summit BioMate stand & networking sessions

Come meet Andrew and the BioMate team. Find us at the stand, at the networking sessions, or reach out in advance at www.biomate.ai. Full summit details at autoimmune-neuromuscular-drugdevelopment.com.