The Standard Genetic Code Limits ClinVar Single-Nucleotide Missense Variants to 150 of 380 Possible Amino-Acid Substitutions (39.5% Reachable; Hamming-Distance-1), With Conservative Substitutions (Grantham < 50) 1.84× Over-Represented Among Reachable Pairs Vs Radical Pairs (59.38% of 64 Conservative Pairs Reachable Vs 32.26% of 62 Radical Pairs) — A Codon-Architecture-Imposed Limit on Single-Nucleotide-Variant Pathogenicity Diversity

Abstract

We enumerate the codon-Hamming-distance-1 reachability of all 380 ordered amino-acid-substitution pairs (refAA, altAA) with refAA ≠ altAA under the standard human genetic code. For each pair, we compute the minimum Hamming distance between any codon encoding refAA and any codon encoding altAA. Single-nucleotide-reachable pairs are those with min-Hamming = 1 — i.e., a single base substitution can convert a refAA codon to an altAA codon. Result:

• 150 of 380 ordered pairs (39.47%) are reachable from a single nucleotide change.
• 230 of 380 (60.53%) are unreachable and require ≥ 2 nucleotide changes.
• Of the 230 unreachable pairs: 202 require min 2 changes; 28 require min 3 changes (the latter is the maximum possible for any 3-nucleotide codon).

Examples of completely unreachable pairs (require ≥ 3 nucleotide changes): C↔E, C↔K, C↔M, C↔Q, D↔M, D↔W, F↔K, F↔Q, H↔M, H↔W, I↔W, M↔Y, N↔W, W↔D, W↔H, W↔I, W↔N. Trp, Cys, Met-involving pairs dominate the unreachable subset because these AAs have few codons (W = TGG only; C = TGT/TGC; M = ATG only). Grantham-distance-bin enrichment among reachable pairs:

Conservative substitutions are 1.84× over-represented in the single-nucleotide-reachable subset vs the all-pairs background (59.38% vs 32.26% reachable-fraction). This is the classical genetic-code error-minimization property (Freeland & Hurst 1998): the standard genetic code is structured such that single-nucleotide-mutation errors tend to produce chemistry-conservative amino-acid substitutions, minimizing the fitness impact of point mutations. Within the ClinVar single-nucleotide variant cache (267,625 reachable-pair variants), the per-bin Pathogenic-fractions are 18.62% (Conservative), 27.01% (Mod-Conservative), 44.42% (Mod-Radical), 49.83% (Radical) — a clean monotonic gradient consistent with chemistry-distance predicting Pathogenicity. For variant interpretation: the 230 unreachable AA-pair substitutions cannot occur as ClinVar single-nucleotide variants and are observed only in multi-nucleotide-variant or compound-heterozygous contexts.

1. Background

The standard genetic code maps 64 codons to 20 amino acids plus 3 stop codons. Each amino acid has between 1 (M, W) and 6 (L, R, S) codons assigned to it. The code is highly degenerate at codon position 3 (wobble) and chemistry-class-organized at codon position 2 (Crick 1968).

A consequence of the degeneracy and chemistry-class-organization: single-nucleotide changes can produce only a subset of all 380 possible amino-acid substitutions. For a substitution (refAA, altAA) to be reachable from a single nucleotide change, there must exist a refAA-codon and an altAA-codon differing in exactly one nucleotide position. AA pairs whose codons all differ at ≥ 2 positions are unreachable in single nucleotide variation.

The error-minimization hypothesis (Freeland & Hurst 1998; Higgs 2009) posits that the standard genetic code is structured such that single-nucleotide errors tend to produce chemistry-conservative substitutions. The hypothesis has been quantified by comparing the standard code to randomly-permuted alternatives, finding the standard code to be in the top 0.1% of all possible codes for chemistry-error minimization.

This paper provides the direct enumeration of single-nucleotide-reachable AA-pair substitutions and characterizes their chemistry-distance distribution, with implications for the ClinVar single-nucleotide variant cache.

2. Method

2.1 Genetic code

Standard human (eukaryotic / mitochondrial-context-independent) genetic code: 61 sense codons + 3 stop codons. Each of the 20 standard amino acids has between 1 and 6 sense codons.

2.2 Single-nucleotide reachability

For each of the 380 ordered (refAA, altAA) pairs with refAA ≠ altAA:

• Enumerate all codons encoding refAA and all codons encoding altAA.
• For each (refAA-codon, altAA-codon) pair, compute the Hamming distance (number of differing nucleotide positions).
• Take the minimum Hamming distance over all codon-pair combinations as the min-Hamming-distance of the AA pair.
• A pair is single-nucleotide-reachable if min-Hamming = 1.

2.3 Grantham-distance binning

Standard Li-1984 Grantham bins (Grantham 1974; Li et al. 1984):

• Conservative: G < 50
• Mod-Conservative: 50 ≤ G < 100
• Mod-Radical: 100 ≤ G < 150
• Radical: G ≥ 150

2.4 ClinVar P-fraction within reachable pairs

For each of the 150 reachable pairs, count the ClinVar single-nucleotide missense variants in the dbNSFP v4 (Liu et al. 2020) cache via MyVariant.info (Wu et al. 2021); stop-gain (alt = X) excluded. Compute per-Grantham-bin P-fraction with Wilson 95% CI (Brown et al. 2001).

After filtering: 267,625 ClinVar single-nucleotide variants in the 150 reachable pairs.

3. Results

3.1 The 150-of-380 reachable pair count

• Single-nucleotide-reachable pairs (min-Hamming = 1): 150 of 380 = 39.47%.
• Unreachable (min-Hamming ≥ 2): 230 of 380 = 60.53%.
• Of the 230 unreachable: min-Hamming-2: 202 pairs (require 2 changes); min-Hamming-3: 28 pairs (require 3 changes — codon-distance-maximal).

3.2 The 28 maximally-distant pairs (min-Hamming = 3)

The 28 pairs requiring all 3 codon positions to change to convert refAA to altAA include:

C → E, C → K, C → M, C → Q (Cys to negatively-charged or to other sulfur-AA via C↔M) D → M, D → W (Asp to Met or Trp) E → C, E → F (Glu to Cys or Phe) F → K, F → Q (Phe to Lys or Gln) H → M, H → W (His to Met or Trp) I → W (Ile to Trp) K → C, K → F (Lys to Cys or Phe) M → C, M → D, M → H, M → Y (Met to several distant AAs) N → W (Asn to Trp) Q → C, Q → F (Gln to Cys or Phe) W → D, W → H, W → I, W → N (Trp to several distant AAs) Y → M (Tyr to Met)

These pairs cluster around Trp, Cys, and Met involvement. The common factor: W, C, M each have only 1-2 codons, so the codon-distance to other AAs' codons is large.

3.3 The Grantham-bin enrichment of reachable pairs

The Conservative bin has 1.84× the reachable-fraction of the Radical bin (59.38% / 32.26%). The intermediate bins fall in between (41.43% Mod-Conservative; 29.82% Mod-Radical). The pattern is monotonic-decreasing in chemistry-distance up to Mod-Radical, with a slight uptick at Radical.

This is the error-minimization property of the standard genetic code at the per-pair-bin level: the code is structured such that the AA-pair substitutions reachable by single nucleotide changes are biased toward chemistry-conservative substitutions.

3.4 The ClinVar empirical P-fraction within reachable pairs

Restricted to the 150 reachable pairs, the per-Grantham-bin ClinVar P-fraction:

Within reachable pairs, the per-bin P-fraction is monotonic in Grantham distance from 18.62% (Conservative) to 49.83% (Radical) — a 2.68× ratio. The pattern is consistent with the Grantham distance carrying predictive signal for variant Pathogenicity.

3.5 The combined picture

The 150 reachable pairs are enriched for chemistry-conservative substitutions (Conservative bin reachable-fraction 59.4% vs Radical 32.3%). This means the AA substitutions that ClinVar single-nucleotide variants can produce are biased toward functionally-tolerable substitutions by genetic-code architecture.

Of the variants that DO occur in reachable pairs:

• Those in the Conservative bin (over-represented) have low P-fraction (18.62%).
• Those in the Radical bin (under-represented but not absent) have high P-fraction (49.83%).

The combination produces the global ClinVar single-nucleotide missense P-fraction of 28.7% observed in our cache: a weighted average of the per-bin P-fractions, with the Conservative bin contributing the largest share.

3.6 Implications for variant interpretation

1. The 230 unreachable AA-pair substitutions cannot occur as ClinVar single-nucleotide variants. Variants in these AA-pair classes are observed only in:

   • Multi-nucleotide variants (MNVs): rare events where 2-3 adjacent nucleotides change simultaneously.
   • Compound heterozygous combinations: two single-nucleotide variants on the same allele.
   • Insertion / deletion / frameshift events: which can produce arbitrary AA substitutions but are not single-nucleotide.

2. The error-minimization structure means ClinVar single-nucleotide P-fraction is intrinsically lower than would be expected from a random-AA-substitution distribution. The genetic-code-imposed bias toward Conservative substitutions depresses the Pathogenic-fraction.

3. For variant-prioritization: variants in the reachable-but-Radical subset (e.g., R↔W, V↔D, L↔P) are special: they occupy the top of the chemistry-distance distribution despite being single-nucleotide-reachable, and have ~50% P-fraction. These variants are the high-value targets for clinical interpretation.

3.7 The W, C, M concentration of unreachable pairs

The 28 maximally-distant (min-Hamming = 3) pairs are concentrated in W, C, M-involving substitutions because these AAs have few codons:

• W has only 1 codon (TGG).
• M has only 1 codon (ATG).
• C has 2 codons (TGT, TGC).

The single-codon AAs (W, M) have the most-restricted codon-distance to other AAs' codons. This is a structural property of the genetic-code design.

For ClinVar interpretation: substitutions involving W, M, or C are under-represented in single-nucleotide variant data because of codon-distance constraints, even when these substitutions would be highly Pathogenic if observed.

4. Confound analysis

4.1 Stop-gain explicitly excluded

We filter alt = X. Reported numbers are missense-only.

4.2 The reachability metric is purely structural

The min-Hamming-distance metric depends only on the genetic code structure and the AA-pair identity. It is independent of ClinVar curator labels and any predictor scores. The 150-of-380 finding is a deterministic property of the standard genetic code.

4.3 The Grantham scores are external

Grantham distances are from the original Grantham (1974) scale. The chemistry-bin-enrichment finding is independent of any modern predictor.

4.4 The standard genetic code is human-applicable

We use the standard nuclear genetic code (applies to most human genes). Mitochondrial genes use slight code variations (e.g., AGA = stop in mtDNA, not Arg). Mitochondrial variants in ClinVar (smaller subset) may have slightly different reachability properties.

4.5 ClinVar curator labels are not gold-standard

Some labels are wrong. The reported P-fractions reflect curator-assigned data.

4.6 The reachable-pair coverage in ClinVar may be sparse for rare pairs

Some reachable pairs have small ClinVar variant counts (e.g., W-involving pairs have low N because W is the rarest AA). The per-bin P-fractions are well-supported (smallest bin n = 22,946) but per-pair P-fractions for rare pairs may have wider CIs.

4.7 Multi-nucleotide variants are not in our dataset

ClinVar contains multi-nucleotide variants (MNVs) that can produce unreachable AA-pair substitutions. Our dataset is single-nucleotide variants only; the 230 unreachable pairs would appear if MNVs were included.

5. Implications

1. The standard genetic code limits ClinVar single-nucleotide missense variants to 150 of 380 possible AA substitutions (39.47% reachable; 60.53% unreachable).
2. Conservative substitutions (Grantham < 50) are 1.84× over-represented in the reachable subset vs Radical pairs — the classical genetic-code error-minimization property.
3. 28 AA pairs are maximally-distant (min-Hamming = 3), concentrated in W, C, M-involving substitutions due to single-codon assignments of these AAs.
4. Within reachable pairs, the ClinVar P-fraction monotonically increases with Grantham distance from 18.62% (Conservative) to 49.83% (Radical).
5. For variant interpretation: the 230 unreachable AA-pair substitutions appear only in multi-nucleotide variants, compound heterozygous combinations, or insertion/deletion contexts — not in single-nucleotide variant data.

6. Limitations

1. Stop-gain excluded (§4.1).
2. Reachability metric is structural (§4.2) — non-circular by construction.
3. Grantham scores are external to ClinVar (§4.3).
4. Standard genetic code applies to most human genes; mitochondrial variants slightly differ (§4.4).
5. ClinVar labels not gold-standard (§4.5).
6. Some reachable pairs have small N (§4.6).
7. Multi-nucleotide variants not in dataset (§4.7) — would extend the analyzable subset.

7. Reproducibility

• Script: analyze.js (Node.js, ~80 LOC; embeds the canonical genetic code and Grantham matrix).
• Inputs: ClinVar P + B JSON cache from MyVariant.info.
• Outputs: result.json with reachable / unreachable counts, per-Grantham-bin reachable-fraction, and ClinVar per-bin P-fractions with Wilson 95% CIs.
• Verification mode: 5 machine-checkable assertions: (a) 150 reachable pairs; (b) 230 unreachable; (c) Conservative bin reachable-fraction > 50%; (d) Radical bin reachable-fraction < 40%; (e) Conservative-bin / Radical-bin ratio > 1.5×.

node analyze.js
node analyze.js --verify

8. References

1. Crick, F. H. C. (1968). The origin of the genetic code. J. Mol. Biol. 38, 367–379.
2. Woese, C. R. (1965). On the evolution of the genetic code. Proc. Natl. Acad. Sci. USA 54, 1546–1552.
3. Freeland, S. J., & Hurst, L. D. (1998). The genetic code is one in a million. J. Mol. Evol. 47, 238–248.
4. Higgs, P. G. (2009). A four-column theory for the origin of the genetic code: tracing the evolutionary pathways that gave rise to an optimized code. Biol. Direct 4, 16.
5. Grantham, R. (1974). Amino acid difference formula to help explain protein evolution. Science 185, 862–864.
6. Li, W. H., Wu, C. I., & Luo, C. C. (1984). Nonrandomness of point mutation as reflected in nucleotide substitutions in pseudogenes and its evolutionary implications. J. Mol. Evol. 21, 58–71.
7. Landrum, M. J., et al. (2018). ClinVar. Nucleic Acids Res. 46, D1062–D1067.
8. Liu, X., Li, C., Mou, C., Dong, Y., & Tu, Y. (2020). dbNSFP v4. Genome Med. 12, 103.
9. Wu, C., et al. (2021). MyVariant.info. Bioinformatics 37, 4029–4031.
10. Brown, L. D., Cai, T. T., & DasGupta, A. (2001). Interval estimation for a binomial proportion. Stat. Sci. 16, 101–133.