mRNA Manufacturability Checker¶

The bridge between computational sequence design and what's actually synthesizable at production scale.

Why this matters¶

A mRNA sequence that's algorithmically perfect can still be impossible to manufacture. Real production requires checking for motifs that destabilize the DNA template, block ribosome scanning, or trigger mRNA decay — issues that show up only in the wet lab.

Eight checks¶

Check	Severity threshold	What it catches
`poly_a_runs`	≥7 As → error	Poly-A runs in the DNA template destabilize the plasmid during IVT
`gc_5prime_hairpin`	≥70% GC → warn, ≥80% → error	GC-rich stems at the 5' end block ribosome scanning
`kozak_strength`	<45% match → warn	Strong Kozak consensus (`GCCRCCATGG`) is required for mammalian translation
`are_motif`	≥1 nonamer → warn, ≥2 → error	AU-rich elements in the 3' UTR trigger mRNA decay
`stop_context`	leakiness ≥0.7 → warn	TGA is the leakiest stop codon; TAA-T or TGA-T is strongest
`hidden_stops`	any internal stop → error	Internal in-frame stops break the protein
`gc_window_uniformity`	stddev >15% → warn	Extreme local GC variation hurts IVT yield
`cpg_balance`	<0.5% or >15% → warn	Suppressed CpG silences; excessive CpG activates TLR9

CLI¶

mrnavax manufacture --cds input.fasta --utr5 GCCGCCACC --utr3 AAAAAAAAAAAAAA

Returns a JSON report with per-check status, score, severity, and summary. Exit code 0 if no errors, 2 if any check returns error.

Python API¶

from mrnavax.manufacturability import score_manufacturability

report = score_manufacturability(
    cds,
    utr5="GCCGCCACC",       # optional, for Kozak scoring
    utr3="AAAAAAAAAAAAAA",  # optional, for ARE detection
)

print(f"overall: {report.overall_score:.3f}")
for c in report.checks:
    print(f"  [{c.severity}] {c.name}: {c.summary}")

Reference¶

Holtkamp et al. (2006) Blood 108.
Kozak (1986) Cell 44.
Chen & Shyu (1995) Trends Biochem Sci 20.