The variant landscape

One premature stop. Three possible routes.

Each variant gets three separate questions: drug readthrough, DNA editing, or exact-restoration suppressor tRNA.

70,376variants with a complete score
6,058mapped condition entities
19exact-restoration tRNA designs
01

Small molecules

A broad prediction narrows when uncertainty bites.

Point estimates screen broadly; interval lower bounds demand stronger support. Decay and residue tolerance narrow both.

Every stronger requirement removes candidates

Threshold sweep across 70,376 scoreable ClinVar nonsense variants

predicted
minimum predicted readthrough≥ 0.5%≥ 1.0%≥ 2.0%Prediction clears the thresholdpale = model midpoint only · dark = interval clears020,00040,00060,00080,00070,269 variants clear the point estimate43,419 variants clear the interval lower bound70,26943,41959,797 variants clear the point estimate15,488 variants clear the interval lower bound59,79715,48827,638 variants clear the point estimate165 variants clear the interval lower bound27,638165candidate variants (farther right = more) · axis runs 0 to 80,000Decay + insertion tolerance also clearpale = model midpoint only · dark = interval clears01,0002,0003,0003,472 variants clear the point estimate1,509 variants clear the interval lower bound3,4721,5092,757 variants clear the point estimate251 variants clear the interval lower bound2,757251505 variants clear the point estimate0 variants clear the interval lower bound5050 · NONEcandidate variants (farther right = more) · axis runs 0 to 3,500
  • pale — model midpoint clears
  • dark — interval lower bound also clears

DARK BARS ARE STRONGER SUPPORT · More candidates means broader predicted reach, not proven benefit.

Details & limits
“All gates” requires predicted readthrough, transcript escape and a majority of tested amino-acid substitutions to be tolerated. It is a candidate screen, not evidence that a compound works in a patient.
251high-confidence candidates

The midpoint alone leaves 2,757—more reach, weaker support.

02

Base editing

Editors can reach nearly a third. Reach is only geometry.

This is editor placement geometry—not efficiency, delivery, off-targets, or clinical eligibility.

Placement reaches 22,042 variants; removing extra nearby edits retains 18,249

Complete partition of 70,660 variants scoreable against BE4max + ABE7.10, SpCas9 NGG

rule
BE4max + ABE7.10, SpCas9 NGGBEST OUTCOME: EXACT REFERENCE AMINO ACID RESTOREDeditor can be placedexact restoration: 7,622 (10.8%)EXACT7,62210.8%alternative restoration: 14,420 (20.4%)ALTERNATIVE14,42020.4%not reachable: 48,618 (68.8%)NO PLACEMENT48,61868.8%and no extra nearby editexact restoration: 6,384 (9.0%)EXACT6,3849.0%alternative restoration: 11,865 (16.8%)ALTERNATIVE11,86516.8%not reachable: 52,411 (74.2%)NO PLACEMENT52,41174.2%green = candidate placement · grey = editor cannot be positioned

GREEN MEANS A CANDIDATE PLACEMENT · Exact restoration is preferable; neither green segment demonstrates efficacy.

Details & limits
Exact means the edit restores the reference amino acid. Alternative means it creates a different sense codon. A bystander-free placement has no additional editable base inside the declared activity window; it still needs experimental efficiency and specificity testing.
31.2%geometrically reachable

7,622 restore the exact amino acid.

03

Transcript survival

Whether the RNA survives depends on which rule you ask.

Adding start-proximal and long-exon exceptions reveals variants whose decay prediction depends on the rule.

The full nonsense-mediated decay rule expands predicted escape from 10.1% to 30.3%

A complete partition of 70,376 scoreable variants

rule
What happens to the RNA before a therapy can act?70,376 scoreable variantslast-exon guideline escape: 7,120 (10.1%)STRICT ESCAPE7,12010.1%added by full rules: 14,231 (20.2%)RULE-SENSITIVE14,23120.2%predicted decay: 49,025 (69.7%)PREDICTED DECAY49,02569.7%RNA REMAINSNEEDS MEASUREMENTRNA REMOVED · READTHROUGH UNLIKELY

GREEN KEEPS THE READTHROUGH ROUTE OPEN · Grey predicts RNA loss; orange marks rule-sensitive uncertainty.

Details & limits
The orange band is not an error bar. It is the exact set whose classification changes when the declared biological exceptions are added. Those 14,231 variants are the most valuable targets for orthogonal transcript-stability measurements.
20.2%rule-sensitive

One in five classifications changes. Both verdicts stay visible.

04

Suppressor tRNA

Two designs reach a third. The tail still matters.

Each design adds only exact-restoration variants not reached by earlier designs.

Nineteen stop-residue designs close the exact-restoration universe

Marginal variants per design, with cumulative coverage over the same scoreable denominator

rule
share of variants with exact restorationDESIGN ID = STOP CODON – RESTORED AMINO ACIDHIGHER BLUE LINE = BROADER EXACT-RESTORATION REACH0.0%25.0%50.0%75.0%100.0%15101519UAG-Q adds 13,550 variants (19.3%)UGA-R adds 10,318 variants (14.7%)UAG-E adds 4,924 variants (7.0%)UAA-E adds 4,658 variants (6.6%)UAG-W adds 4,554 variants (6.5%)UGA-W adds 4,504 variants (6.4%)UAA-Q adds 4,354 variants (6.2%)UAG-Y adds 4,163 variants (5.9%)UAA-Y adds 3,994 variants (5.7%)UGA-C adds 2,662 variants (3.8%)UGA-S adds 2,291 variants (3.3%)UAG-K adds 1,937 variants (2.8%)UAA-K adds 1,775 variants (2.5%)UGA-G adds 1,746 variants (2.5%)UAA-S adds 1,393 variants (2.0%)UGA-L adds 1,014 variants (1.4%)UAG-S adds 931 variants (1.3%)UAG-L adds 917 variants (1.3%)UAA-L adds 691 variants (1.0%)1. UAG-Q: 13,550 variants (19.3%)2. UGA-R: 23,868 variants (33.9%)3. UAG-E: 28,792 variants (40.9%)4. UAA-E: 33,450 variants (47.5%)5. UAG-W: 38,004 variants (54.0%)6. UGA-W: 42,508 variants (60.4%)7. UAA-Q: 46,862 variants (66.6%)8. UAG-Y: 51,025 variants (72.5%)9. UAA-Y: 55,019 variants (78.2%)10. UGA-C: 57,681 variants (82.0%)11. UGA-S: 59,972 variants (85.2%)12. UAG-K: 61,909 variants (88.0%)13. UAA-K: 63,684 variants (90.5%)14. UGA-G: 65,430 variants (93.0%)15. UAA-S: 66,823 variants (95.0%)16. UGA-L: 67,837 variants (96.4%)17. UAG-S: 68,768 variants (97.7%)18. UAG-L: 69,685 variants (99.0%)19. UAA-L: 70,376 variants (100.0%)UAG-Q · 19.3%UGA-R · 33.9%UAA-L · 100.0%designs added in frontier order
  • new variants added by this design
  • cumulative exact-restoration coverage

Coverage is sequence logic, not therapeutic evidence. No suppressor tRNA in this project has a qualifying ribosome-profiling measurement.

Details & limits
The first design, UAG-Q, reaches 13,550 variants. UGA-R takes the pair to 23,868. The final designs add few variants, but they are the only exact-restoration route for those stop-residue combinations. UAG-Q means a design for UAG stops that restores glutamine (Q).
33.9%covered by two designs

Best starting portfolio, not evidence that either construct works or is safe.