Skip to content

Methods and checksMethods

How every result is reached and what it was checked against, app by app. The same as each app's Methods dialog.

Methods and checks pages

Drybench Protein: methods and checks

How Drybench Protein reaches its results: every calculator, analysis or view with its methods, the references behind them, how far each has been checked, and the app's limits.

Open Drybench Protein · What the levels mean · Every app's methods and checks

Retained structure evidence

Capture versions: 4.0.0 for new experimental structure images and 3.0.0 for new predicted-model images; historical records retain their recorded procedure.

Its result goes through 1 method. The lowest level among them: Tested, not reference-checked.

Below the top level: Exact predicted-model image and receipt capture (Tested, not reference-checked).

Methods

Exact predicted-model image and receipt capture

  • Level: Tested, not reference-checked
  • How it was checked: Automated tests cover the capture's records and refusals, and its exact paired runs are compared with the 0.6.0 reference values; the Mol* image itself has no reference, and the byte, PNG and scene references were recorded for earlier capture versions. Integrity checks do not establish biological validity or structural sequence verification.

How it is made and tested

  • Capture versions and historical records: Version 4.0.0 applies to new experimental structure captures and 3.0.0 to new predicted-model captures. Historical records retain their recorded procedure and original mapping policy; opening this overview does not upgrade them.
  • Retained image and receipt integrity, actual source identity, finite native camera values, capture limits, field-named refusals and text-projection parity. Independent Python fixtures supply byte hashes, PNG boundaries, scene identities and the 3IKA point substitution; browser checks cover actual rendering separately.
  • New predicted schema2/adapter2/capture3 receipts preserve exact paired runs, source confidence and actual loaded-file identity. Source-independent rereading rejects mixed versions, contradictory statuses, altered hashes, bridged holes and pending correspondence; historical v1 semantics and lazy/full projection parity remain strict.

Experimental structure evidence

Loaded-file correspondence and actual viewer application remain separate facts in each new experimental receipt.

Its result goes through 1 method. The lowest level among them: Tested, not reference-checked.

Below the top level: Loaded-file experimental image and receipt capture (Tested, not reference-checked).

Methods

Loaded-file experimental image and receipt capture

  • Level: Tested, not reference-checked
  • How it was checked: Automated tests compare the scientific inner and application consistency rules with verification 0.7.0, republished readably as 0.10.0. Those application cases are synthetic relations, not renderer observations. Native focus and capture require separate browser checks; the Mol* image has no independent reference. Integrity checks do not establish biological validity.

How it is made and tested

  • Capture versions and historical records: Version 4.0.0 applies to new experimental structure captures and 3.0.0 to new predicted-model captures. Historical records retain their recorded procedure and original mapping policy; opening this overview does not upgrade them.
  • Retained image and receipt integrity, actual source identity, finite native camera values, capture limits, field-named refusals and text-projection parity. Independent Python fixtures supply byte hashes, PNG boundaries, scene identities and the 3IKA point substitution; browser checks cover actual rendering separately.

Source confidence and exact residue transfer

Source-model colours and current-UniProt confidence have separate coordinates and coverage.

Its result goes through 2 methods. The lowest level among them: Checked against a reference.

Methods

Invariant standard-residue correspondence

  • Level: Checked against a reference
  • Reference: Gotoh O. An improved algorithm for matching biological sequences. J Mol Biol. 1982;162(3):705–708. doi:10.1016/0022-2836(82)90398-9
  • How it was checked: Matched exactly, on recorded cases, the expected values of two separate Python references: Biopython 1.88 listing every optimal alignment (47 cases, verification 0.5.1) and an exact-integer path-counting script written from the method text (8 cases, including inputs above 20,000 optimal alignments and three real AlphaFold DB/UniProt pairs, verification 0.8.0). Both references were written by the app's authors. Agreement holds for this alignment policy only; it does not establish biological equivalence, and inputs outside the recorded cases are not shown.

Source confidence parsing and exact transfer

  • Level: Checked against a reference
  • Reference: Varadi M, et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res. 2022;50(D1):D439–D444. doi:10.1093/nar/gkab1061
  • How it was checked: Matched 102 recorded cases from a separate Python script (Biopython 1.88, verification 0.6.0): parser outcomes, bands, missing positions, counts, viewer targets and source binding exactly, means within 1e-12. The script applies the app's stated cut-off rule; it was not compared with the provider's own category letters. Source-model confidence does not validate the current sequence.

How it is made and tested

  • The frozen independent Biopython 1.88 reference supplies strict source parsing, exact-only transfer, precision at band edges, sparse and zero scores, contributing-count means, split focus targets and source identity/domain refusals. Recorded EGFR assets establish representative full-model coordinates, without claiming runtime structural sequence verification.
  • Worker cancellation during hashing and execution; exactly-once teardown after success, timeout, malformed/stale response, worker, constructor or posting failure; bounded closed messages, actual sequence/confidence bindings and separate legacy-only admission. No main-thread alignment fallback.
  • Exact confidence transfer: Source-model scores stay in model coordinates. Current-UniProt confidence and predicted focus use only identical standard residues paired in every optimum of the fixed global/free-end BLOSUM62 alignment, gap cost 10 + 0.5(k − 1). Other positions stay missing; range means state contributing n over selected N. Unverified fragment domains are withheld. The mapping runs in a cancellable worker with a 10-second operational limit; each sequence is at most 10,000 residues and the length product at most 10 million.

Dossier

The protein's UniProt entry, its variants, drugs and AlphaFold model as each source records them, with the retrieval time on every chip and the source's release where it reports one.

No method record is kept for this view, so it carries no level.

How it is made and tested

  • Accession formats and isoform suffixes, search queries (including gene symbols shaped like accessions, such as P2RY12), the ranking of suggestions, curated variants and the reading of a UniProt entry.
  • COSMIC-only records are dropped, clinical classes are ranked, variants are counted per residue and named in UniProt numbering (HBB residues 7 and 8 among the cases).
  • Open Targets drug parsing: salt forms merged, stages ranked, approvals that rest only on an ATC code not counted (efaproxiral), and withdrawals by market (voxelotor, ketorolac, metamizole, troglitazone).
  • Choosing the AlphaFold DB model, placing a model built for an older UniProt sequence (RAB12-, SLC3A2- and POLR2A-like cases), the pLDDT cut-offs, the boundary rule, and AlphaMissense averaging.
  • pLDDT bands: AlphaFold DB's cut-offs of 90, 70 and 50 (Varadi et al. 2022, read 24 October 2026); a score exactly on a cut-off counts in the higher band, read from each residue's score. Some viewers, including RCSB PDB's model pages, place a score of exactly 90 or 70 in the lower band.
  • Withdrawn from the market: Only when an Open Targets drug warning is worldwide, or covers both the United States and the European Union.

Compare

The protein's PDB entries and AlphaFold model side by side, with the default entry chosen by the residues it resolves and its resolution.

No method record is kept for this view, so it carries no level.

How it is made and tested

  • PDBe best structures and UniPDB ranges, the resolution weighting, the default-entry ranking, SIFTS residue mapping, and recorded cases: the K-Ras4B stretch in 5TAR, an engineered point mutation, and 6GOD at exactly 95% identity.
  • The MolViewSpec scenes: one colour per pLDDT band, unresolved residues faded, the chosen chain highlighted, and local copies of structure files.
  • Default PDB entry: The entry that resolves the most residues of the protein, weighted by resolution: full weight to 2 Å, half at 4 Å, less for coarser maps and NMR.
  • Sequence identity: A leading candidate at 95% SIFTS sequence identity or less counts for half.
  • Large entries: Entries over 150,000 atoms load only the chosen chain until you ask for the whole entry.

Check a substitution

One amino-acid substitution checked against the current UniProt sequence, with the AlphaMissense score only where the AlphaFold model's sequence matches it.

No method record is kept for this view, so it carries no level.

How it is made and tested

  • The substitution check against a Python oracle's write-once fixtures (BRAF V600E in five notations, V599E refused, EGFR L858R and T790M, mature-numbered L834R refused, models offset either way, withheld entries, every refusal); each score also found in AlphaMissense's source file; every score's text as the file writes it.

Compare with another entry

Two UniProt sequences aligned by Gotoh's affine-gap dynamic programming with BLOSUM62, a gap of k residues costing 10 + 0.5(k − 1); global with end gaps free unless you choose end gaps counted, or local.

No method record is kept for this view, so it carries no level.

How it is made and tested

  • The aligner against Biopython 1.88's PairwiseAligner on UniProt inputs (TP53 human and mouse, TP53 and its isoform 2, a tagged EGFR kinase construct, a one-residue insertion, identical sequences, B, Z and X codes, short peptides), runs and repeats with tied best alignments, and gaps in opposite sequences side by side at an end, in three modes: the optimal score exactly, the shown alignment re-scored to it, the identity, similarity and gap counts, the number of co-optimal alignments and the residue pairs shared by all of them; refusals over 10 million cells and for selenocysteine; BLOSUM62 entry by entry.

Data sources

Limits

Not medical advice. Nothing here is advice about any medicine, treatment or condition, and nothing here is an assessment of any person's variant, sample or molecule. Decisions made with it are yours.

Not a forecast. Stages, scores, rates, rankings and similarities shown here describe records as they stand, or historical averages; they are not a forecast for any molecule, trial, drug, company or target.

  • Not medical advice. Variant classifications are what UniProt, ClinVar or Ensembl report, not an assessment of anyone's variant. AlphaMissense scores are predictions that have not been validated for clinical use and can disagree with known disease variants.
  • Current-UniProt pLDDT and predicted focus use only identical standard residues paired in every optimum of the fixed sequence alignment. Missing sequence, unverified model fragments, ambiguous correspondence and missing scores withhold values. Source-model colours remain separate; AlphaMissense keeps its conservative legacy placement.
  • The default PDB entry is picked by a simple rule (residues resolved, weighted by resolution). It may be a truncated construct, carry engineered mutations or lack the ligand you care about, so check its title and use the list to choose another.
  • The experimental view shows the deposited model, not necessarily the biological assembly. Entries over 150,000 atoms load only the chosen chain until you ask for the whole entry.
  • The predicted model is a single-chain prediction of the full UniProt sequence, signal peptides and propeptides included. It has no ligands, partners or modifications, low-pLDDT regions should not be read as real structure, and AlphaFold has not been validated for predicting the effect of mutations. PAE is not shown here, so the relative placement of domains cannot be judged from this page.
  • AlphaFold DB does not supply a model for every protein. Source-model coordinates belong to its release's sequence; exact correspondence does not establish biological equivalence. Isoform requests open the canonical entry with a note and source link. Mapping is bounded at 10,000 residues per sequence and a 10-million length product.
  • Drug data covers human targets only and lists molecules whose curated mechanism of action names the target. Stages are the highest Open Targets reports for any indication and region, not an approval for every indication shown, and not a forecast for any trial. Withdrawals come from Open Targets' drug warnings. One that is not worldwide and does not cover both the United States and the European Union is shown with its countries and does not by itself move a drug with a regulator's approval record out of Approved; other withdrawal reports are shown as reports for one product or indication.
  • All data is live. If a provider is slow or down, that panel says so and offers a retry; the rest of the page still works.