Can public data confirm charging and misinsertion? — assessment 2026-10-30

Q1. IS THE tRNA CHARGED WITH ALANINE?  -> NO PUBLIC DATA CAN ANSWER THIS.
  Periodate charging-state tRNA-seq exists for human cells (e.g. GEO GSE188861),
  but it reports only WHETHER a tRNA is aminoacylated, never WHICH amino acid is attached.
  No high-throughput public assay identifies the amino acid on a named tRNA.
  Requires: in vitro aminoacylation with AlaRS, or MS of the aminoacyl-tRNA.
  Prior evidence (strong but not a measurement): Hou & Schimmel 1988 - inserting G3:U70 into
  unrelated tRNAs confers alanine acceptance; Miller 1991 - G3:C70 abolishes it.

Q2. DOES IT INSERT Ala AT Val CODONS?  -> YES IN PRINCIPLE, VIA A GENOTYPE CONTRAST.
  Single untargeted proteome: UNDERPOWERED.
    Tennakoon overexpressed construct gives 5e-2 misincorporation at Val codons.
    Scaling to this gene's observed abundance predicts ~5e-5 to 5e-4.
    Background translation error rate ~1e-4 => signal at or below noise.
  Genotype-stratified design (carriers vs non-carriers) cancels the background:
    detect +5e-4  -> ~22,000 Val observations per genotype group
    detect +1e-4  -> ~235,000 per group
    detect +5e-5  -> ~784,000 per group
    (two-proportion test, alpha 0.05, power 80%)
  Scale available: ~1e6 PSMs per deep proteome; Val ~6% of residues; peptides ~15 aa
    => ~1e6 Val observations per deep dataset. CPTAC (deep proteomes + matched WGS)
       aggregates far more, so even the pessimistic 5e-5 target is reachable.
  Protocol: split donors by rs1233572 genotype; re-search raw files allowing the
    Val->Ala mass shift of -28.0313 Da at Val positions; compare per-donor rates.
  Blocker here: requires tens of GB of raw files and a full search-engine run.
