MyoScore Calculator

Run MyoScore directly in your browser — no server, no installation. Upload or paste a bulk RNA-seq raw count matrix and compute per-sample scores across five genetically-driven muscle-health dimensions.

Input format

• First column: gene identifier (Gene Symbol or Ensembl ID)
• First row: sample names
• Body: raw counts (not TPM/FPKM)
• Separator: CSV or TSV (auto-detected)

gene_symbol,Healthy_01,Healthy_02,Myopathy_01,Myopathy_02
NFS1,530,621,410,380
NEK4,812,904,650,702
RFT1,443,510,298,315
...

For Ensembl IDs (e.g. ENSG00000100300), the calculator lazy-loads a GTEx-derived annotation dictionary (~56 000 entries, 1.5 MB) and converts them to Gene Symbols automatically. Version suffixes like .15 are stripped before lookup.

How it works

The calculation mirrors the MyoScore R package:

1. Normalization — raw counts are scaled to log₂(CPM + 1).
2. Gene selection — expression vectors are subset to the 591 weighted gene-dimension entries (417 unique Gene Symbols).
3. Z-score standardization — each gene is centered and scaled across all input samples.
4. Weighted aggregation — z-scores are multiplied by GWAS/TWAS direction (±1) and TWAS-derived weight, then averaged per dimension.
5. Min-max scaling — raw dimension scores are rescaled to 0–100.
6. Composite score — the five dimensions are combined via a GWAS-derived weighted mean:

$$\text{MyoScore}=0.252\cdot \text{Strength}+0.177\cdot \text{Mass}+0.243\cdot \text{LeanMuscle}+0.242\cdot \text{Youth}+0.087\cdot \text{Resilience}$$

All computation happens client-side — your count matrix never leaves the browser.

Recommended cohort size

Min-max scaling is cohort-relative, so results depend on the distribution of your input samples:

Samples	Behavior
1	Not supported (at least 2 required).
2–19	Scores are produced but unstable — a warning is displayed.
≥ 20	Recommended.

For benchmarking an individual against a reference, pool your sample with a healthy cohort before scoring, or use the R package with the reference transform.

Privacy

• No data is uploaded to any server.
• The calculator runs entirely as static JavaScript hosted on Cloudflare Pages.
• The gene weight table and annotation dictionary are static assets fetched directly by your browser.

See also

• Methods — full derivation of weights and directions
• R package — command-line & batch workflow
• Validation — MRI, single-cell, and clinical benchmarks