Reference checks

Reference checks: Drybench Sequence

This page lists the reference checks published for Drybench Sequence. Each set gives the methods it covers, what the expected values were computed with, the number of cases, the tolerance the app must meet, the date, and the file holding the inputs and expected outputs, so you can open a case and compare it with what the app gives.

Every case was chosen and recorded by the Drybench team; none has been independently reviewed. A reference check tests the calculation, not your data, your experiment or whether a model suits it. Methods without a set are listed on the methods page with their level.

Methods and levels Documentation

Published sets

Published verification sets

Methods	Oracle	Fixtures	Tolerance	Date	File
primer-design-bounded-flanks, oligo-tm-nearest-neighbour	Biopython	34	Tm within 1e-6 degrees; retained qualifications, normalized requests and literal record claims exact	2026-10-20	primer-design-qualification-v4.jsonRecorded for version 0.9.0; the test suite checks the current engine against it.
primer-design-bounded-flanks	Biopython	23	asymmetric requested/retained correction agreement and historical records exact	2026-10-11	primer-qualification-effective-salt-v1.jsonRecorded for version 0.6.0; the test suite checks the current engine against it.
translation-ncbi-genetic-codes	Biopython	113	exact	2026-10-06	translation.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
orf-finder-ncbi-conventions	NCBI ORFfinder	18	exact, every field of every list	2026-10-06	orf.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
orf-finder-ncbi-conventions	Circular ORF derivation	64	exact, every field of every list, on the record and on every listed rotation	2026-10-14	orf-circular-v1.jsonRecorded for version 0.9.0; the test suite checks the current engine against it.
oligo-tm-nearest-neighbour, hairpin-dimer-dg37	Buffer-admissibility derivation	20	verdict, rule and structure exact; monovalent and free Mg²⁺ within 1e-9 mM	2026-10-12	buffer-admissibility-v1.jsonRecorded for version 0.9.0; the test suite checks the current engine against it.
protein-average-mass, isoelectric-point-bjellqvist, extinction-280-pace	Expasy ProtParam	22	MW within 0.01 Da, pI within 0.01, ε280 exact	2026-10-06	protein.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
oligo-tm-nearest-neighbour, salt-correction-owczarzy-2004, salt-correction-owczarzy-2008, salt-correction-santalucia-1998	Biopython	49	within 0.1 °C under both concentration models; regime, every fitted-range note, refusals and acceptance exact	2026-10-20	oligo-tm-v2.jsonRecorded for version 0.9.0; the test suite checks the current engine against it.
hairpin-dimer-dg37	primer3-py	15	within 0.4 kcal/mol on ntthal's own structures (8 cases here and 7 salt probes in hairpin-dimer-probes.json)	2026-10-06	hairpin-dimer.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
hairpin-dimer-dg37	primer3-py	16	within 0.1 kcal/mol on ntthal's own structures (12 cases here and 4 salt probes in hairpin-dimer-probes.json)	2026-10-06	hairpin-dimer.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
nucleic-acid-average-mass	Python fractions	5	within 0.01 g/mol	2026-10-06	dna-mw.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
pcr-product-exact-pair	Exact-match derivation	30	exact, refusals included	2026-10-10	pcr-product.jsonRecorded for version 0.3.0; the test suite checks the current engine against it.
exact-fragment-search	Exact-match derivation	26	exact, refusals included	2026-10-10	fragment-search.jsonRecorded for version 0.3.0; the test suite checks the current engine against it.
primer-design-bounded-flanks	Primer-design derivation	13	every returned pair, coordinate, product, count and cap exact; Tm and ranking terms within 1e-6 °C; refusals exact	2026-10-11	primer-design.jsonRecorded for version 0.5.0; the test suite checks the current engine against it.
sdm-primers-two-schemes, oligo-tm-nearest-neighbour	Mutagenesis-primer derivation	18	every primer sequence, coordinate, matched flank, count by reason, cap and refusal exact; GC within 1e-9; every melting temperature within 1e-6 °C	2026-10-17	sdm-primers-v1.jsonRecorded for version 0.7.0; the test suite checks the current engine against it.
exact-transform-global-edit-v2	Revision-transform derivation	10	kind, map, hashes, withheld state and annotation finding exact; edit count equals the independent edit distance	2026-10-10	revision-transforms.jsonRecorded for version 0.4.0; the test suite checks the current engine against it.
dna-align-v1	Biopython	60	the optimum on both strands, the co-optimal count, the common pairs, the differences and the placement groups exact; the alignment shown is the published tie order's first	2026-10-20	dna-align-v1.jsonRecorded for version 0.8.0; the test suite checks the current engine against it.
read-check-v1	Biopython	44	kept ranges, states, strands, scores, counts, listings, totals and junctions exact; the peak ratio as an exact fraction on one trace (RC-18); at the settings the set covers only	2026-10-20	pipeline/sequence-toolkit/fixtures/readcheck/readcheck-v1.jsonRecorded for version 0.10.0; the test suite checks the current engine against it.
peptide-smiles-linear	RDKit	101	the same canonical SMILES on 89 peptides; the 12 refusals exact	2026-10-06	peptide-smiles.jsonRecorded for version 0.1.0; the test suite checks the current engine against it.
codon-adaptation-sharp-li-v1	Codon adaptation derivation	30	CAI within 1e-9; the scored count, every left-out count, the replaced cells, the rare list and every withholding exact	2026-11-07	pipeline/sequence-toolkit/fixtures/codon-usage/cai-v1.jsonRecorded for version 0.11.0; the test suite checks the current engine against it.
dna-msa-v1	Multiple alignment derivation	19	the alignment, merge order, scores and column counts exact; refusals exact	2026-11-07	pipeline/sequence-toolkit/fixtures/msa/msa-v1.jsonRecorded for version 0.12.0; the test suite checks the current engine against it.
protein-msa-v1	Protein multiple alignment derivation	23	the alignment, merge order, scores, column counts and zero-scoring letters exact; refusals exact	2026-11-07	pipeline/sequence-toolkit/fixtures/protein-msa/protein-msa-v1.json
feature-annotation	Annotation suggestion derivation	3	every placement's entry, start, end, strand and mismatches exact, identity within 1e-12, and no other match call; partial calls and gapped alignments not covered	2026-11-07	pipeline/sequence-toolkit/fixtures/annotation/annotation-v1.jsonRecorded for version 0.14.0; the test suite checks the current engine against it.
codon-recode-most-used-v1	Codon recoding derivation	5	the recoded letters, the protein, the codon and base counts, the left-as-they-were list and every constraint count exact	2026-11-07	pipeline/sequence-toolkit/fixtures/codon-usage/recode-v1.jsonRecorded for version 0.11.0; the test suite checks the current engine against it.

Oracles and build-time tools

Software the app's numbers were checked against, or its data built with. None of it ships with the app or is a dependency of it.

Biopython 1.88 (Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-06). frame translations and the oligo Tm fixtures (MeltingTemp.Tm_NN with the named tables and corrections)
Exact-match derivation exact_matches.py with Biopython 1.88 (MIT (the script); Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-10). the PCR product and fragment search fixtures, every window tested directly, one case worked by hand
Primer-design derivation primer_design.py with Biopython 1.88 (MIT (the script); Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-11). the bounded primer-design fixtures: every melting temperature from MeltingTemp.Tm_NN, the candidate windows, counts, ranking and caps derived in the script, exact binding by the exact-match derivation, one case worked by hand
Mutagenesis-primer derivation sdm_primers.py with Biopython 1.88 (MIT (the script); Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-17). the site-directed mutagenesis primer fixtures: the mutant sequence, the candidate windows, the matched flanks, GC, the ranking and the counts by reason derived in the script, every melting temperature from MeltingTemp.Tm_NN and every reverse complement from Bio.Seq, one case per scheme worked by hand
Revision-transform derivation revision-transforms.py with Biopython 1.88 (MIT (the script); Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-10). the design-revision review fixtures: hand-worked coordinate maps and an independent edit distance
primer3-py 2.3.1 (GPL-2.0-or-later; Oracle; 2026-10-06). hairpin, self-dimer and cross-dimer ΔG at 37 °C (libprimer3 2.6.1 thal, run offline as ntthal); never shipped and its parameter files never copied
NCBI ORFfinder 0.4.3 (Public domain (NCBI); Oracle; 2026-10-06). cross-checking the ORF fixtures
Circular ORF derivation orf_circular.py with Biopython 1.88 (MIT (the script); Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-14). the circular ORF fixtures: each strand read round the origin by the rules on the methods page, as seven copies read linearly and as a modular walk, agreeing exactly; Biopython translations; every listed rotation; the linear scan and NCBI ORFfinder 0.4.3 on each unrolled reading, and the linear scan on each ORF inside the numbering
Buffer-admissibility derivation buffer_admissibility.py with Biopython 1.88 (MIT (the script); Biopython License Agreement (BSD-3-Clause file by file); Oracle; 2026-10-12). the buffer verdicts shared by the Tm and the structure checks: the admissibility bounds restated, free Mg²⁺ by bisection, and MeltingTemp.Tm_NN on every admitted buffer
Expasy ProtParam web form, responses saved 2026-10-06 (SIB terms of use (To be checked); Oracle; 2026-10-06). molecular weight, isoelectric point and ε280 of 22 proteins and peptides
RDKit 2026.03.6 (BSD-3-Clause; Oracle; 2026-10-06). the canonical SMILES of 89 peptides (MolFromSequence) for the molecule hand-off
Python fractions 3.11.15 (Python-2.0; Oracle; 2026-10-06). exact arithmetic for the DNA molecular weight fixtures
Codon adaptation derivation cai_reference.py (Python fractions and decimal) (MIT (the script); Oracle; 2026-11-07). the codon adaptation index fixtures: every w a fraction, the product exact, the root in decimal at 60 digits, on 15 hand cases (the three-codon case primary) and three real coding sequences (E. coli lacZ, yeast ADH1 and human INS) against each of the five host tables; cross-checked against Biopython 1.88's CodonAdaptationIndex
Codon recount codon_recount.py (MIT (the script); Oracle; 2026-11-07). the second, independent count of each host's codon-usage table from the same RefSeq files, by a different route from the build script
Multiple alignment derivation msa_reference.py (MIT (the script); Oracle; 2026-11-07). the progressive multiple alignment fixtures, the method written apart from the engine
Protein multiple alignment derivation protein_msa_reference.py (MIT (the script); Oracle; 2026-11-07). the progressive multiple alignment fixtures for proteins: the method written apart from the engine, BLOSUM62 typed from NCBI's file with the file's hash asserted
MAFFT 7.505 (BSD-style; Oracle; 2026-11-07). an offline comparison on four DNA and four protein multiple alignment sets (--nuc and --amino), recorded, never the gate
Annotation suggestion derivation annotation_reference.py (MIT (the script); Oracle; 2026-11-07). the auto-annotation suggestions' hand-checkable set: every ungapped full-length placement of a library entry within the match threshold on pUC19 (as stated and read across the origin) and pBR322, found by shared words and checked letter by letter
Codon recoding derivation recode_reference.py (MIT (the script); Oracle; 2026-11-07). the recoding fixtures: a second implementation of the recoding rules with the finished-sequence constraint re-check written apart from the repair, the protein asserted identical before and after

Fixture tables, computed when the page was built

Every row is one fixture compared with the engine when this page was built; the full sets are the published files, which pipeline/sequence-toolkit/reference.py, peptides.py and exact_matches.py wrote and whose hashes the manifest records. No value in a file comes from the engine under test.

Frame translations in genetic codes 1, 2 and 11

Expected values from Biopython 1.88 (Bio.Seq.translate), recorded 2026-10-06; tolerance exact; 12 of 108 shown, all passing. translation.json sha256 2f4debf7f26cdcf9…

Input	Expected	Ours	Difference	Pass
ins-mrna.t1.f+1.full	155 aa, SPPGQAASEEAI…	155 aa, SPPGQAASEEAI…	identical	yes
ins-mrna.t1.f+1.to_stop	45 aa, SPPGQAASEEAI…	45 aa, SPPGQAASEEAI…	identical	yes
ins-mrna.t1.f+2.full	154 aa, ALQDRLHQKRPS…	154 aa, ALQDRLHQKRPS…	identical	yes
ins-mrna.t1.f+2.to_stop	38 aa, ALQDRLHQKRPS…	38 aa, ALQDRLHQKRPS…	identical	yes
ins-mrna.t1.f+3.full	154 aa, PSRTGCIRRGHQ…	154 aa, PSRTGCIRRGHQ…	identical	yes
ins-mrna.t1.f+3.to_stop	129 aa, PSRTGCIRRGHQ…	129 aa, PSRTGCIRRGHQ…	identical	yes
ins-mrna.t1.f-1.full	155 aa, AGSRALFHLSRC…	155 aa, AGSRALFHLSRC…	identical	yes
ins-mrna.t1.f-1.to_stop	147 aa, AGSRALFHLSRC…	147 aa, AGSRALFHLSRC…	identical	yes
ins-mrna.t1.f-2.full	154 aa, LVQGLYSISLGA…	154 aa, LVQGLYSISLGA…	identical	yes
ins-mrna.t1.f-2.to_stop	94 aa, LVQGLYSISLGA…	94 aa, LVQGLYSISLGA…	identical	yes
ins-mrna.t1.f-3.full	154 aa, WFKGFIPSLSVQ…	154 aa, WFKGFIPSLSVQ…	identical	yes
ins-mrna.t1.f-3.to_stop	24 aa, WFKGFIPSLSVQ…	24 aa, WFKGFIPSLSVQ…	identical	yes

Open reading frames, every list field by field

Expected values from an independent scan cross-checked with Biopython and NCBI ORFfinder 0.4.3, recorded 2026-10-06; tolerance exact; 12 of 18 shown, all passing. orf.json sha256 631f0a2853fe0219…

Input	Expected	Ours	Difference	Pass
ins-mrna.t1.atgOnly	1 listed, 1 nested, 1 open	1 listed, 1 nested, 1 open	every field equal	yes
ins-mrna.t1.alternativeStarts	2 listed, 6 nested, 1 open	2 listed, 6 nested, 1 open	every field equal	yes
ins-mrna.t2.atgOnly	1 listed, 0 nested, 0 open	1 listed, 0 nested, 0 open	every field equal	yes
ins-mrna.t2.alternativeStarts	3 listed, 1 nested, 0 open	3 listed, 1 nested, 0 open	every field equal	yes
ins-mrna.t11.atgOnly	1 listed, 1 nested, 1 open	1 listed, 1 nested, 1 open	every field equal	yes
ins-mrna.t11.alternativeStarts	3 listed, 4 nested, 2 open	3 listed, 4 nested, 2 open	every field equal	yes
mt-atp8-atp6-cox3.t1.atgOnly	8 listed, 2 nested, 0 open	8 listed, 2 nested, 0 open	every field equal	yes
mt-atp8-atp6-cox3.t1.alternativeStarts	9 listed, 2 nested, 1 open	9 listed, 2 nested, 1 open	every field equal	yes
mt-atp8-atp6-cox3.t2.atgOnly	2 listed, 4 nested, 1 open	2 listed, 4 nested, 1 open	every field equal	yes
mt-atp8-atp6-cox3.t2.alternativeStarts	3 listed, 7 nested, 1 open	3 listed, 7 nested, 1 open	every field equal	yes
mt-atp8-atp6-cox3.t11.atgOnly	8 listed, 2 nested, 0 open	8 listed, 2 nested, 0 open	every field equal	yes
mt-atp8-atp6-cox3.t11.alternativeStarts	9 listed, 11 nested, 1 open	9 listed, 11 nested, 1 open	every field equal	yes

Molecular weight, pI and ε280 of 22 proteins and peptides

Expected values from Expasy ProtParam (saved responses), recorded 2026-10-06; tolerance MW within 0.01 Da, pI within 0.01, ε280 exact; 12 of 22 shown, all passing. protein.json sha256 de4163c56cf4b5d9…

Input	Expected	Ours	Difference	Pass
P01 Human preproinsulin (signal, B chain, C peptide, A chain) (110 aa)	11980.91 Da, pI 5.22, ε 17335/16960	11980.91 Da, pI 5.22, ε 17335/16960	MW -0.001, pI +0.00	yes
P02 Aequorea victoria green fluorescent protein (238 aa)	26886.32 Da, pI 5.67, ε 22015/21890	26886.32 Da, pI 5.67, ε 22015/21890	MW +0.001, pI +0.00	yes
P03 Human ubiquitin (first unit of polyubiquitin-C) (76 aa)	8564.84 Da, pI 6.56, ε 1490/1490	8564.84 Da, pI 6.56, ε 1490/1490	MW +0.005, pI +0.00	yes
P04 Hen egg-white lysozyme C, mature chain (129 aa)	14313.14 Da, pI 9.32, ε 37970/37470	14313.14 Da, pI 9.32, ε 37970/37470	MW -0.001, pI +0.00	yes
P05 Bovine serum albumin, mature chain (583 aa)	66432.96 Da, pI 5.60, ε 42925/40800	66432.96 Da, pI 5.60, ε 42925/40800	MW -0.005, pI +0.00	yes
P06 Human metallothionein-2 (61 aa)	6042.16 Da, pI 8.23, ε 1250/0	6042.16 Da, pI 8.23, ε 1250/0	MW +0.005, pI +0.00	yes
P07 Human thymosin beta-4, mature chain (43 aa)	4921.46 Da, pI 5.02, ε 0/0	4921.46 Da, pI 5.02, ε 0/0	MW +0.002, pI +0.00	yes
P08 Human prothymosin alpha, N-terminally processed chain (110 aa)	12071.77 Da, pI 3.66, ε 0/0	12071.77 Da, pI 3.66, ε 0/0	MW -0.001, pI +0.00	yes
P09 Human macrophage migration inhibitory factor, mature chain (114 aa)	12345.11 Da, pI 8.24, ε 13075/12950	12345.11 Da, pI 8.24, ε 13075/12950	MW -0.003, pI +0.00	yes
P10 Human hemoglobin subunit alpha, mature chain (141 aa)	15126.36 Da, pI 8.73, ε 9970/9970	15126.36 Da, pI 8.73, ε 9970/9970	MW -0.004, pI +0.00	yes
P11 Human myoglobin, mature chain (153 aa)	17052.61 Da, pI 7.29, ε 13980/13980	17052.61 Da, pI 7.29, ε 13980/13980	MW +0.004, pI +0.00	yes
P12 Human cytochrome c, mature chain (104 aa)	11617.53 Da, pI 9.59, ε 13075/12950	11617.53 Da, pI 9.59, ε 13075/12950	MW +0.002, pI +0.00	yes

Oligo Tm under both concentration models

Expected values from Biopython 1.88 (MeltingTemp.Tm_NN), recorded 2026-10-06; tolerance within 0.1 °C; 12 of 45 shown, all passing. oligo-tm.json sha256 1588ca109f2e31df…

Input	Expected	Ours	Difference	Pass
T01 d15-at (50 mM Na⁺)	31.08 / 28.66 °C	31.08 / 28.66 °C	+0.000 / +0.000	yes
T02 d15-gc (50 mM Na⁺)	65.33 / 62.76 °C	65.33 / 62.76 °C	+0.000 / +0.000	yes
T03 M13F-20 (50 mM Na⁺)	53.58 / 51.40 °C	53.58 / 51.40 °C	+0.000 / +0.000	yes
T04 T7-promoter (50 mM Na⁺)	47.60 / 45.70 °C	47.60 / 45.70 °C	+0.000 / +0.000	yes
T05 T7-promoter (50 mM Na⁺)	48.55 / 46.65 °C	48.55 / 46.65 °C	+0.000 / +0.000	yes
T06 T7-promoter (1000 mM Na⁺)	63.95 / 61.86 °C	63.95 / 61.86 °C	+0.000 / +0.000	yes
T07 SP6 (50 mM Na⁺)	42.77 / 40.68 °C	42.77 / 40.68 °C	+0.000 / +0.000	yes
T08 sc16 (50 mM Na⁺)	55.14 / 53.97 °C	55.14 / 53.97 °C	+0.000 / +0.000	yes
T09 d30-at (1000 mM Na⁺)	72.05 / 70.60 °C	72.05 / 70.60 °C	-0.000 / -0.000	yes
T10 d35 (50 mM Na⁺)	76.60 / 75.43 °C	76.60 / 75.43 °C	+0.000 / +0.000	yes
T11 d40-at (50 mM Na⁺)	51.96 / 51.00 °C	51.96 / 51.00 °C	+0.000 / +0.000	yes
T12 d40-gc (1000 mM Na⁺)	99.98 / 98.88 °C	99.98 / 98.88 °C	+0.000 / +0.000	yes

DNA molecular weight

Expected values from Python 3.11.15 fractions (hand arithmetic), recorded 2026-10-06; tolerance within 0.01 g/mol; 5 of 5 shown, all passing. dna-mw.json sha256 15f60ea1db40d55d…

Input	Expected	Ours	Difference	Pass
M01 dA (one nucleotide)	251.246 g/mol	251.246 g/mol	+0.0000	yes
M02 T7 promoter primer	6125.053 g/mol	6125.053 g/mol	+0.0000	yes
M03 M13 forward (-20) primer	5228.467 g/mol	5228.467 g/mol	+0.0000	yes
M04 oligo(dT)20	6021.936 g/mol	6021.936 g/mol	+0.0000	yes
M05 designed 40-mer	12320.042 g/mol	12320.042 g/mol	+0.0000	yes

Predicted PCR products

Expected values from an independent derivation (exact_matches.py, Biopython 1.88), recorded 2026-10-10; tolerance exact; 12 of 30 shown, all passing. pcr-product.json sha256 dd145d143e6f6542…

Input	Expected	Ours	Difference	Pass
pcr-hand-unique	unique, 1 product(s), first 25 bp	unique, 1 product(s), first 25 bp	identical	yes
pcr-two-forward-sites	ambiguous, 2 product(s), first 37 bp	ambiguous, 2 product(s), first 37 bp	identical	yes
pcr-reverse-template	unique, 1 product(s), first 25 bp	unique, 1 product(s), first 25 bp	identical	yes
pcr-circular-origin	unique, 1 product(s), first 25 bp	unique, 1 product(s), first 25 bp	identical	yes
pcr-linear-no-wrap	none, 0 product(s)	none, 0 product(s)	identical	yes
pcr-site-across-origin	unique, 1 product(s), first 25 bp	unique, 1 product(s), first 25 bp	identical	yes
pcr-palindromic-pair	ambiguous, 2 product(s), first 22 bp	ambiguous, 2 product(s), first 22 bp	identical	yes
pcr-tails	unique, 1 product(s), first 41 bp	unique, 1 product(s), first 41 bp	identical	yes
pcr-tail-not-inferred	none, 0 product(s)	none, 0 product(s)	identical	yes
pcr-normalised-input	unique, 1 product(s), first 25 bp	unique, 1 product(s), first 25 bp	identical	yes
pcr-overlap-only	refused: overlapping-sites	refused: overlapping-sites	identical	yes
pcr-adjacent	unique, 1 product(s), first 20 bp	unique, 1 product(s), first 20 bp	identical	yes

Exact fragment search

Expected values from an independent derivation (exact_matches.py, Biopython 1.88), recorded 2026-10-10; tolerance exact; 12 of 26 shown, all passing. fragment-search.json sha256 dffa9bebd1843720…

Input	Expected	Ours	Difference	Pass
frag-forward	3-8 +	3-8 +	identical	yes
frag-reverse	3-8 -	3-8 -	identical	yes
frag-palindrome-once	3-8 both; 11-16 both	3-8 both; 11-16 both	identical	yes
frag-homopolymer-overlaps	2-5 +; 3-6 +; 4-7 +; 5-8 +	2-5 +; 3-6 +; 4-7 +; 5-8 +	identical	yes
frag-homopolymer-reverse	2-5 -; 3-6 -; 4-7 -	2-5 -; 3-6 -; 4-7 -	identical	yes
frag-tandem-repeat	1-8 +; 6-13 +; 11-18 +	1-8 +; 6-13 +; 11-18 +	identical	yes
frag-iupac-barrier	6-11 +	6-11 +	identical	yes
frag-iupac-codes	19-24 +	19-24 +	identical	yes
frag-linear-no-wrap	no hit	no hit	identical	yes
frag-circular-origin	7-4 + wraps	7-4 + wraps	identical	yes
frag-circular-origin-reverse	7-4 - wraps	7-4 - wraps	identical	yes
frag-circular-full-length	3-2 + wraps	3-2 + wraps	identical	yes

Peptides as SMILES

Expected values from RDKit 2026.03.6 (MolFromSequence), recorded 2026-10-06; tolerance the same canonical SMILES, each read back through RDKit.js; refusals exact; 12 of 101 shown, all passing. peptide-smiles.json sha256 4ca03c17768ce37a…

Input	Expected	Ours	Difference	Pass
N-A (3 residues)	C[C@H](N)C(=O)NCC(=O)N[C@@H](C)C(=O)O	C[C@H](N)C(=O)NCC(=O)N[C@@H](C)C(=O)O	the same canonical SMILES	yes
M-A (3 residues)	C[C@H](NC(=O)[C@H](C)NC(=O)CN)C(=O)O	C[C@H](NC(=O)[C@H](C)NC(=O)CN)C(=O)O	the same canonical SMILES	yes
C-A (3 residues)	C[C@H](NC(=O)[C@H](C)NC(=O)CN)C(=O)O	C[C@H](NC(=O)[C@H](C)NC(=O)CN)C(=O)O	the same canonical SMILES	yes
1-A (1 residues)	C[C@H](N)C(=O)O	C[C@H](N)C(=O)O	the same canonical SMILES	yes
N-C (3 residues)	C[C@H](NC(=O)CNC(=O)[C@@H](N)CS)C(=O)O	C[C@H](NC(=O)CNC(=O)[C@@H](N)CS)C(=O)O	the same canonical SMILES	yes
M-C (3 residues)	C[C@H](NC(=O)[C@H](CS)NC(=O)CN)C(=O)O	C[C@H](NC(=O)[C@H](CS)NC(=O)CN)C(=O)O	the same canonical SMILES	yes
C-C (3 residues)	C[C@H](NC(=O)CN)C(=O)N[C@@H](CS)C(=O)O	C[C@H](NC(=O)CN)C(=O)N[C@@H](CS)C(=O)O	the same canonical SMILES	yes
1-C (1 residues)	N[C@@H](CS)C(=O)O	N[C@@H](CS)C(=O)O	the same canonical SMILES	yes
N-D (3 residues)	C[C@H](NC(=O)CNC(=O)[C@@H](N)CC(=O)O)C(=…	C[C@H](NC(=O)CNC(=O)[C@@H](N)CC(=O)O)C(=…	the same canonical SMILES	yes
M-D (3 residues)	C[C@H](NC(=O)[C@H](CC(=O)O)NC(=O)CN)C(=O…	C[C@H](NC(=O)[C@H](CC(=O)O)NC(=O)CN)C(=O…	the same canonical SMILES	yes
C-D (3 residues)	C[C@H](NC(=O)CN)C(=O)N[C@@H](CC(=O)O)C(=…	C[C@H](NC(=O)CN)C(=O)N[C@@H](CC(=O)O)C(=…	the same canonical SMILES	yes
1-D (1 residues)	N[C@@H](CC(=O)O)C(=O)O	N[C@@H](CC(=O)O)C(=O)O	the same canonical SMILES	yes

Hairpins against the oracle, case by case

ntthal 2.6.1 (primer3-py 2.3.1), recorded 2026-10-06: 15 fixtures (15 in the gate), the oracle's own structure rescored under this engine's rules within 0.4 kcal/mol on every case with a structure and this engine's ΔG at or below that rescore; all passing. The two structures coincide on 6 of 13; the oracle reports the structure with the highest melting temperature, this engine the lowest ΔG°37. The badge reads the same numbers: within 0.4 kcal/mol, 15 fixtures.

Case	Sequences	Oracle ΔG	Rescore of its structure	Gap and cause	Our ΔG	Our structure	Coincide	Pass
H01 hairpin	CCCCCATCCGATCAGGGGG	-3.18	-2.99	0.19one salt step: the oracle counts one pair fewer for a stem reaching the oligo's end	-2.99	/////---------\\\\\	yes	yes
H02 hairpin	GCGCGAAAAAACGCGC	-4.79	-4.42	0.37one salt step: the oracle counts one pair fewer for a stem reaching the oligo's end	-4.42	/////------\\\\\	yes	yes
H03 hairpin	ATATATTTTTTATATAT	none			none		yes	yes
H04 hairpin	ACTGCAGGCTTGCGATTTTCGCAAGCC	-9.21	-8.99	0.21one salt step: the oracle counts one pair fewer for a stem reaching the oligo's end	-9.63	------////////-----\\\\\\\\	no	yes
H05 hairpin	TTTTTTTTTTTTTTTTTTTT	none			none		yes	yes
H06 hairpin	GAATTTATTGAAAACAATAAATTC	-3.62	-3.37	0.25one salt step: the oracle counts one pair fewer for a stem reaching the oligo's end	-5.25	//////////----\\\\\\\\\\	no	yes
H07 hairpin	GGATCCGAATTCGGATCC	-5.12	-5.09	0.04	-5.09	///////----\\\\\\\	yes	yes
H08 hairpin	TTGACCGAGTGCAACGTACGGATCTGTTGCACTCGGTCAATGAC	-11.42	-11.19	0.23the complementary-flank rule: the oracle keeps a single dangling end at a helix end	-14.36	///////////////----------\\\\\\\\\\\\\\\----	no	yes
S01 hairpin	TTGACCGAGTGCAACGTACGGATCTGTTGCACTCGGTCAATGAC	-7.91	-7.68	0.23the complementary-flank rule: the oracle keeps a single dangling end at a helix end	-9.89	///////////////----------\\\\\\\\\\\\\\\----	no	yes
S02 hairpin	TTGACCGAGTGCAACGTACGGATCTGTTGCACTCGGTCAATGAC	-9.93	-9.70	0.23the complementary-flank rule: the oracle keeps a single dangling end at a helix end	-12.46	///////////////----------\\\\\\\\\\\\\\\----	no	yes
S03 hairpin	TTGACCGAGTGCAACGTACGGATCTGTTGCACTCGGTCAATGAC	-11.31	-11.08	0.23the complementary-flank rule: the oracle keeps a single dangling end at a helix end	-14.22	///////////////----------\\\\\\\\\\\\\\\----	no	yes
S04 hairpin	TTGACCGAGTGCAACGTACGGATCTGTTGCACTCGGTCAATGAC	-13.69	-13.46	0.23the complementary-flank rule: the oracle keeps a single dangling end at a helix end	-17.25	///////////////----------\\\\\\\\\\\\\\\----	no	yes
S09 hairpin	TGCGCGAAAAAACGCGCT	-5.31	-5.32	0.02	-5.32	-/////------\\\\\-	yes	yes
S10 hairpin	GCGCGAAAAAACGCGC	-5.82	-5.79	0.02	-5.79	/////------\\\\\	yes	yes
S11 hairpin	TGCGCGAAAAAACGCGCT	-6.67	-6.69	0.02	-6.69	-/////------\\\\\-	yes	yes

Dimers against the oracle, case by case

ntthal 2.6.1 (primer3-py 2.3.1), recorded 2026-10-06: 16 fixtures (16 in the gate), the oracle's own structure rescored under this engine's rules within 0.1 kcal/mol on every case with a structure and this engine's ΔG at or below that rescore; all passing. The two structures coincide on 13 of 14; the oracle reports the structure with the highest melting temperature, this engine the lowest ΔG°37. The badge reads the same numbers: within 0.1 kcal/mol, 16 fixtures.

Case	Sequences	Oracle ΔG	Rescore of its structure	Gap and cause	Our ΔG	Our structure	Coincide	Pass
D01 self-dimer	GCGAATTCGCATGCAAGCTT	-12.06	-12.14	0.07	-12.14	ATGCAAGCTT GCGAATTCGC CGCTTAAGCG TTCGAACGTA ----------	yes	yes
D02 self-dimer	TTGCCAGTCAAGTCAGCATGC	-6.45	-6.42	0.04	-6.42	TTGCCAGTCAAGTCA --------------- GCATGC CGTACG ACTGAACTGACCGTT	yes	yes
D03 self-dimer	AAAAAAAAAAAAAAAAAAAA	none			none		yes	yes
D04 self-dimer	CTGACGTAAGCTTACGTCAG	-21.11	-21.11	0.00	-21.11	CTGACGTAAGCTTACGTCAG GACTGCATTCGAATGCAGTC	yes	yes
D05 self-dimer	TAATACGACTCACTATAGGG	-3.80	-3.76	0.05	-3.76	TAATACGACTCA GG---------- CTATAG GATATC GG ACTCAGCATAAT	yes	yes
D06 self-dimer	GGCCATTAGTCAGTACGGCC	-3.56	-3.55	0.00	-5.03	ATTAGTCAGTACGGCC GGCC CCGG CCGGCATGACTGATTA ----------------	no	yes
X01 cross-dimer	TAATACGACTCACTATAGGG / CCCTATAGTGAGTCGTATTA	-17.31	-17.31	0.00	-17.31	TAATACGACTCACTATAGGG ATTATGCTGAGTGATATCCC	yes	yes
X02 cross-dimer	AGCTTGACCTGAAGGCATGC / TCAGTCCATGGATGGCATGC	-6.30	-6.38	0.08	-6.38	AGCTTGACCTGAAG -------------- GCATGC CGTACG GTAGGTACCTGACT	yes	yes
X03 cross-dimer	AAAAAAAAAAAAAAAAAAAA / CCCCCCCCCCCCCCCCCCCC	none			none		yes	yes
X04 cross-dimer	GTAAAACGACGGCCAGT / CAGGAAACAGCTATGAC	-2.68	-2.69	0.01	-2.69	GTAAAACGACGG AGT CC GG CAGTATCGACAAA AC-	yes	yes
X05 cross-dimer	CGCAAATGGGCGGTAGGCGTG / TAGAAGGCACAGTCGAGG	-2.71	-2.73	0.02	-2.73	CGCAAA G GGTAGGCGTG TG GC AC CG GGAGCTG A GAAGAT----	yes	yes
X06 cross-dimer	TAATACGACTCACTATAGGG / CCCTATAGTAAGTCGTATTA	-12.49	-12.49	0.00	-12.49	C TAATACGACT ACTATAGGG ATTATGCTGA TGATATCCC A	yes	yes
S05 self-dimer	TTGCCAGTCAAGTCAGCATGC	-4.86	-4.82	0.04	-4.82	TTGCCAGTCAAGTCA --------------- GCATGC CGTACG ACTGAACTGACCGTT	yes	yes
S06 self-dimer	TTGCCAGTCAAGTCAGCATGC	-5.78	-5.74	0.04	-5.74	TTGCCAGTCAAGTCA --------------- GCATGC CGTACG ACTGAACTGACCGTT	yes	yes
S07 self-dimer	TTGCCAGTCAAGTCAGCATGC	-6.40	-6.37	0.04	-6.37	TTGCCAGTCAAGTCA --------------- GCATGC CGTACG ACTGAACTGACCGTT	yes	yes
S08 self-dimer	TTGCCAGTCAAGTCAGCATGC	-7.49	-7.45	0.04	-7.45	TTGCCAGTCAAGTCA --------------- GCATGC CGTACG ACTGAACTGACCGTT	yes	yes