Names and identifiers

GL1X does not define a nomenclature of its own. It is an index: every structure, chain, residue, ligand, protein and modification it shows is keyed by an identifier that a public authority assigns (the wwPDB, UniProt, the IUBMB, Unimod, IUPAC), and every name it shows is copied from the structure file or from the record's metadata. Where GL1X has to decide something the files do not state, such as whether glycerol in a crystal is an additive or a ligand, which amino acid a selenomethionine stands for in a sequence, or what kind of bond holds a sugar to an asparagine, it uses small lookup tables keyed by those same identifiers.

This page answers three questions: which identifier names each kind of thing, where each name GL1X shows comes from, and which tables GL1X keeps of its own. The identifiers table and every lookup table below are written from GL1X's code: a test compares them with the tables the program uses and fails when they differ, so the page cannot fall behind the program. The API serves the same tables, with a version that changes whenever any of them does.

In short #

• Keys are public identifiers. A ligand is its wwPDB Chemical Component Dictionary code (HEM, ZN, NAG), a structure its PDB ID, a protein its UniProtKB accession, an enzyme activity its EC number, a modification in a spectrum its Unimod name. RCSB PDB, PDBe, UniProt and BioLiP use the same keys, so anything found in GL1X can be looked up there under the same key, and GL1X's pages link to those resources by it.
• Names come from the files. A component's name, synonyms and formula are the ones the structure file gives, which the PDB fills from the CCD when it releases an entry; a molecule's name is the file's description of its entity; a record's title is the one its MolNFT metadata gives. GL1X holds no copy of the CCD or of UniProt and asks neither while it runs.
• GL1X's own tables are classes, not names. They sort what files hold into the groups GL1X's tools need: five ligand classes, the parent amino acid or base of each residue name it reads, a description, category and Unimod name for common modified residues, the kinds of group bonded to a residue, and the groups, charges, environments and residue classes of the pH analysis. They rename nothing: a component missing from a table keeps its code and its name.
• One alphabet is GL1X's own: GSA, the 20-letter structural alphabet that structure search runs on, described with its training in Structure alignment, GSA and structure search.

The identifiers #

MolNFT records and their sequences #

A MolNFT record is an NFT, and an NFT is identified on its chain by two numbers: the address of the collection's contract and the token id within it. GL1X keys every record by that pair, in its pages (/token/{contract}/instance/{tokenId}, /bio/workbench/{contract}/{tokenId}), its API (/api/bio/records/{contract}/{tokenId}/…) and its datasets, because nothing else is guaranteed to be unique: two records can hold the same PDB entry (one in MolNFT v1 and one in v2, for instance), and a record's scientific identifier is whatever its minter wrote. That scientific identifier, the record's molecular id, is its PDB ID when it has one, otherwise its accession, read from the record's metadata; when the metadata gives neither, GL1X takes the PDB ID and the accession the record's own structure file states.

A record can hold several sequences, and each has a sequence key: the key the record's metadata gives it, or, when it gives none, chain: and the chain's identifier, or sequence: and the sequence's place in the list; the record's main sequence, when it is not one of those, is primary. Keys are at most 128 characters of letters, digits and . _ : -, and a repeated key gets -2, -3… BLAST hits, datasets and proofs name a sequence by its record and key, together with the SHA-256 digest of its letters, which is what a proof checks (see MolNFT: sequences, keys and digests).

PDB IDs #

The wwPDB gives every entry of the Protein Data Bank a PDB ID of four characters, a digit followed by three letters or digits (1AKI, 4DST). IDs are not case-sensitive; GL1X writes them in capitals. The four-character space is filling up, and the wwPDB has introduced extended PDB IDs of twelve characters, pdb_ followed by eight letters or digits: an existing entry's extended ID is its ID with pdb_0000 in front (1aki becomes pdb_00001aki), and from 21 July 2027 newly deposited entries will receive extended IDs only (wwPDB: extended PDB IDs). GL1X recognizes both forms in a record's metadata. A structure is linked to its RCSB PDB and PDBe pages by its ID.

A predicted model has no PDB ID. AlphaFold DB names its models AF- + the UniProtKB accession + -F + the fragment number (AF-P00698-F1). GL1X reads such names as accessions, and a predicted model appears on a protein's page when its record or its structure file names that protein's UniProtKB accession.

Chains and residue numbers #

A structure file identifies chains and residues twice. The author identifiers (auth_asym_id, auth_seq_id, and the insertion code pdbx_PDB_ins_code) are the ones the depositors chose and the ones papers cite: lysozyme's catalytic glutamate is Glu35 because the authors numbered it so. The label identifiers (label_asym_id, label_seq_id) are assigned by the PDB: each molecule gets its own label chain (a ligand or the waters too), and residues are numbered 1, 2, 3… along the sequence of their entity.

GL1X shows author identifiers everywhere, and falls back to the label ones only when a file lacks them; PDB-format files carry only the author ones (the chain in column 22, the number in columns 23–26, the insertion code in column 27). Author numbers need not start at 1, can be negative (a tag before the mature protein), can skip (a loop that was not modelled) and can repeat with an insertion code: antibodies numbered in the Kabat scheme have residues 100, 100A, 100B, and the serine proteases are numbered after chymotrypsin, so thrombin has a residue 60A. GL1X writes such a residue 100A. Where a tool works on positions in a sequence, such as the motif search, it gives the position in the sequence, counted from 1, next to the author's residue number.

Chemical components and the CCD #

Every distinct residue and small molecule in the PDB, from alanine and adenosine to a drug or a crystallization buffer, has an entry in the wwPDB Chemical Component Dictionary (Westbrook et al. 2015), identified by a code: ALA, A, DA, HEM, ATP, GOL, ZN. The entry describes the chemistry once, for every structure that holds it: its name and synonyms, formula and formal charge, each atom with its name and element, each bond with its order, the stereochemistry, ideal coordinates, and descriptors such as SMILES and InChI. Structure files refer to components by code, so the code is the key a ligand, a modified residue or a nucleotide has across the whole archive.

Codes have been one to three letters and digits. As the three-character space ran out, the wwPDB began issuing five-character codes, and entries that use them are distributed only in PDBx/mmCIF (and PDBML), because the PDB format has no room for them (PDBj: extension of CCD codes). GL1X reads codes of up to five characters from mmCIF and BinaryCIF files.

A code is an identifier, not a name and not a class. HEM is heme b in every structure that holds it; whether it is bound to a globin or a cytochrome is told by the structure, not the code. The CCD types components by their place in a polymer (L-peptide linking, RNA linking, non-polymer, saccharide…), not by their role, so glycerol and heme are both non-polymer; that is why GL1X keeps the ligand-class lists below.

UniProtKB accessions #

UniProt identifies each protein sequence entry by an accession: six characters (P00698) or, for newer entries, ten (A0A023GPI8), in a fixed pattern of letters and digits. The accession is the stable key; the entry name (LYSC_CHICK) can change, and an isoform is the accession with a suffix (P00698-2). GL1X accepts exactly the pattern UniProt documents,

[OPQ][0-9][A-Z0-9]{3}[0-9]   or   [A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2}

and maps a structure's chains to accessions from the file's own database references: DBREF (and the DBREF1/DBREF2 pairs that hold ten-character accessions) in PDB files, _struct_ref and _struct_ref_seq in mmCIF and BinaryCIF, with the database named UNP (or UNIPROT, UNIPROTKB, SWS, TREMBL). A protein page (/bio/protein/P00698) gathers every record with a chain mapped to the accession and every record named after it.

EC numbers #

An enzyme activity is named by its EC number, four fields from the IUBMB's Enzyme Nomenclature: class, subclass, sub-subclass and serial number. Lysozyme, which hydrolyses glycosidic bonds (class 3, hydrolases; 3.2, glycosylases; 3.2.1, those acting on O-glycosyl bonds), is 3.2.1.17. A dash stands for a field not assigned, so 3.4.21.- is some serine endopeptidase. GL1X takes EC numbers from the file: _entity.pdbx_ec in mmCIF, the EC: field of COMPND in PDB files. It reads lists separated by commas, semicolons, spaces or slashes, drops an EC prefix, and reads a number with fewer than four fields as ending in dashes (3.4.21 is 3.4.21.-). Annotation transfer compares EC numbers field by field (see EC numbers by depth).

One-letter codes #

Sequences are written with one-letter codes. Those of IUPAC-IUB (1984) cover the twenty amino acids, B and Z for an unresolved pair (Asp or Asn, Glu or Gln) and X for an unknown residue; U for selenocysteine was added by the IUBMB and IUPAC nomenclature committees in 1999, and O for pyrrolysine and J for an unresolved Leu or Ile came later, as UniProt and the NCBI write them. Nucleic acids use A, C, G, T and U, and N for any base (NC-IUB 1985). In a sequence GL1X writes a modified residue with the letter of the amino acid or base it comes from, so selenomethionine (MSE) is M and pseudouridine (PSU) is U; a D-amino acid is written with the letter of its L form; inosine is N. Which residue names are read as which letter is the table Residues read as amino acids below; a residue of any other name in a protein chain is X.

Unimod names #

Mass spectrometry names a modification by the mass it adds, and the community database of those is Unimod (Creasy and Cottrell 2004): each modification has an accession, a name and an elemental composition, such as Phospho (accession 21, HO₃P, +79.966 Da), Oxidation (35, O) and Carbamidomethyl (4, H₃C₂NO). GL1X's mass-spectrometry tools name modifications by these names and accessions (see Modifications and how to write them). The table of modified residues below gives, for each structure component it knows, the Unimod name of the same chemical change, so that a phosphoserine (SEP) seen in a crystal and a Phospho on serine seen in a spectrum are recognizably the same thing.

The correspondence is chemical, not biological. Selenomethionine has the composition change Unimod calls Delta:S(-1)Se(1), but in a crystal it is almost always engineered, introduced to solve the phase problem; a cysteine carrying 2-mercaptoethanol (CME) is an adduct of the buffer. The category column says which is which: GL1X files such residues as engineered and artifact, not as modifications the cell made.

Atomic weights and isotope masses #

Two kinds of mass appear in GL1X. A standard atomic weight is the mean mass of an element's atoms in normal terrestrial material, weighted by the abundance of its isotopes; IUPAC's Commission on Isotopic Abundances and Atomic Weights (CIAAW) publishes them (Prohaska et al. 2022). Where natural variation makes the standard weight an interval (hydrogen's is [1.00784, 1.00811]), IUPAC also gives a single conventional value (1.008), and those are the values GL1X uses for the formula weight of a component when the file states none (see Weights). A monoisotopic mass is the mass of the molecule built from each element's most abundant isotope, and it is what a mass spectrometer resolves for a peptide; GL1X's proteomics tools use isotope masses from the AME 2016 evaluation (Wang et al. 2017) with IUPAC's representative isotopic abundances (see Monoisotopic and average mass).

DSSP states #

Secondary structure is written in the states of DSSP (Kabsch and Sander 1983), which GL1X assigns from backbone hydrogen bonds as DSSP 4 does: H α-helix, G 3₁₀-helix, I π-helix, P polyproline II helix, E strand in a β-ladder, B isolated β-bridge, T hydrogen-bonded turn, S bend, - none of these, and ? for a residue whose backbone atoms are missing (see Secondary structure (DSSP)). Tools that need three states read H, G and I as helix, E and B as strand, and the rest as coil.

GSA letters #

GSA, the GL1X structural alphabet, is the one alphabet GL1X defines. It writes each residue of a chain as one of 20 states, a to t (lower case, so they cannot be mistaken for amino acids), computed from the Cα trace around the residue, so that structure search can compare chains as strings before TM-align confirms the best candidates. Its states, its substitution matrix and its training are documented in GSA, the GL1X structural alphabet. GSA is versioned (GSA-1); a new version would be a new alphabet, never a silent change.

Groups of the pH analysis #

The pH analysis names each ionizable group by its chain, the residue's name as the file writes it and its author number with any insertion code: A:GLU35 is the side-chain carboxyl of Glu35 in chain A (a histidine a simulation file calls HID is A:HID15), and A:NTERM1 and A:CTERM129 are the free terminal amine and carboxyl of chain A, numbered by the residues that carry them. The kinds are the seven side chains that can titrate (Asp, Glu, His, Cys, Tyr, Lys and Arg, though arginine stays protonated up to about pH 13) and the two termini; the table Ionizable groups of the pH analysis below gives each with its model pKa and its source, and pH environments explains how they are found. Answers carry the model's version (gl1x-ph-1), so a result can be told from one computed with other parameters.

Where the names come from #

GL1X copies names; it changes only their form. It trims quotes and runs of spaces, writes formulas one way (element symbols in their usual case, the charge last: C34 H32 Fe N4 O4, Zn +2), and computes weights from formulas. It does not translate, merge or correct names, and it does not fill in a name a file leaves out: a component that no file names is shown by its code. When files disagree, the catalog keeps the first name a file gave, except that a name in capitals gives way to the same name in mixed case.

How GL1X sorts what a file holds #

Each structure file is read the same way, whatever its format, and the tables below are applied in this order.

1. Polymer or not. mmCIF and BinaryCIF state each entity's type (polymer, non-polymer, branched, water). PDB files do not, so GL1X decides from the records: an ATOM residue belongs to its chain, and so does a HETATM residue that is read as an amino acid or a nucleotide and is bonded to the residue before or after it, or numbered within the span of the chain's ATOM residues (a modified residue written as a heteroatom group). Water (HOH, WAT, DOD, H2O) is set aside, and so are the dummy atoms (DUM) with which the OPM database marks membrane planes.
2. A residue in a chain is an amino acid when its name is in the amino-acid table and it has a Cα atom, and is written with that table's letter; it is a nucleotide when its name is in the nucleotide table and it has a C3′ atom. A nucleotide is DNA when its name is one of DA, DC, DG, DT, DU, DI, DN, RNA when its name is a single letter or it has a 2′-hydroxyl, DNA when it has C1′ or C4′ but no 2′-hydroxyl (5CM, 8OG), RNA otherwise. A residue of any other name keeps its place in the chain: it is an amino acid written X when it has the backbone atoms N, Cα and C (or a Cα alone in an ATOM record or a polymer entity), and a nucleotide written N when it has C3′ together with C1′ or C4′, and a P or O3′ atom (or belongs to a polymer entity). So an unfamiliar modified residue never breaks a chain or shifts its numbering; only its letter is unknown.
3. A non-polymer group is a ligand, and gets one of five classes from the rules in Ligand classes below (metal, ion, additive, sugar, ligand), as the ligand index explains in Classes.
4. A non-standard residue in a protein chain is reported among the structure's observed modifications with the description and category of the modified-residue table (whose Unimod column names the same chemical change for the mass-spectrometry tools); a D-amino acid as a D-amino acid; any other as nonstandard, by its code (see Observed in structures). In a nucleic acid chain it is a nucleotide modification, by its code.
5. A group bonded to a residue (two atoms within the sum of their covalent radii plus 0.45 Å) is named by the first rule that fits: a sugar is a glycan, its category set by the residue and atom it is attached to; a component on the lipid list is a lipid anchor, one on the cofactor list a cofactor; an additive is an additive adduct; anything else a covalent ligand (see Groups bonded to residues). Metals and ions are not read as bonded groups: their contacts are coordination, which the binding sites and metal-site templates describe.
6. The pH analysis finds the titrating groups by residue name, including the protonation-state names that simulation programs write (HID, HIE, HIP, ASH, GLH, LYN…); leaves out a cysteine in a disulfide, a group bonded to a ligand and a terminus capped by a blocking group; and fixes the charge of a group coordinating a metal (see Charges that do not titrate).

Other tables GL1X keeps #

The tables on this page name and class what structure files hold. Other pages list the constants of their own methods in full, next to the method that uses them:

Two sets of pK values appear in GL1X, and they answer different questions. The Bjellqvist values give the isoelectric point of a sequence, as gels and ExPASy do, with every group at its standard pK; the pH analysis uses model pKa values measured on short peptides (Thurlkill et al. 2006; arginine's from Fitch et al. 2015) and shifts each by its place in the structure. A buried aspartate has the same pK in the first and a different one in the second.

The tables #

Everything below is written from GL1X's code by bio.NomenclatureMarkdown, and go test ./internal/bio fails when this part of the page and the code differ. Component codes are the CCD's; names in the modified-residue table are the CCD names in lower case.

Ligand classes #

A ligand of one atom of a metal element is a metal (79 elements):

Ac Ag Al Am As Au Ba Be Bi Bk Ca Cd Ce Cf Cm Co Cr Cs Cu Dy Er Es Eu Fe Fm Fr Ga Gd Ge Hf Hg Ho In Ir K La Li Lu Mg Mn Mo Na Nb Nd Ni Np Os Pa Pb Pd Pm Po Pr Pt Pu Ra Rb Re Rh Ru Sb Sc Sm Sn Sr Ta Tb Tc Th Ti Tl Tm U V W Y Yb Zn Zr

A ligand of one atom of any other element is an ion (unless its component is an additive below), and so is any ligand whose component is one of these inorganic ions and small molecules (33 components):

2HP AZI BO3 BO4 BR CL CMO CO2 CO3 CYN F H2S IOD MOO N2O NH4 NO NO2 NO3 O OH OXY PER PI PO3 PO4 SCN SE4 SO3 SO4 UNX VO4 WO4

These crystallization, buffer, cryoprotectant and detergent components are additives (85 components):

12P 144 15P 1BO 1PE 2PE 7PE ACT ACY B3P BME BOG BTB BU1 BU2 BU3 C8E CAC CIT CXS DD9 DEC DIO DMF DMS DMU DTD DTT DTU EDO EGL EOH EPE ETF FLC FMT GOL HED HEX HEZ HTG IMD IPA LDA LMT MES MLA MLI MOH MPD MPO MRD MXE NHE OCT OLC P33 P4C P6G PDO PE3 PE4 PE5 PE8 PEG PG0 PG4 PG5 PG6 PGE PGO PGR POL PPI SDS SGM SIN TAR TBU TFA TLA TMA TRS UND XPE

A component whose type names a saccharide, a group of a branched entity, or one of these saccharides (for PDB files, which state no type) is a sugar (54 components):

0MK 9MR A2G ADA AHR ARA BDP BEM BG6 BGC BMA BXY CBI FCA FCB FRU FUC FUL G6P GAL GCS GCU GLA GLC GLO GMH IDS KDN KDO LAK LAT LXC MAG MAL MAN MLB MMA NAA NAG NDG NGA NGC NGZ RAM RHA SGN SIA SLB SUC TRE XYP XYS Z9N ZDC

Every other non-polymer group is a ligand. These components are water and are left out (4):

DOD H2O HOH WAT

These are pseudo-atoms, points that are not atoms (the dummy atoms with which the OPM database marks membrane planes), and are left out too:

DUM

Residues read as amino acids #

A residue with one of these names is an amino acid of a protein chain, written with the one-letter code of the amino acid it is or comes from: modified residues as their parent, D-amino acids as their L form (92 names).

Residues read as nucleotides #

A residue with one of these names is a nucleotide, written with its parent base; inosine and unknown nucleotides are N (43 names).

Modified amino acids #

The observed modifications of a structure describe a modified residue by its component, with what it is and its category; the Unimod column names the same chemical change where one fits, as the mass-spectrometry tools write it (47 components; any other non-standard residue is nonstandard).

D-amino acids #

These components are D-amino acids (19), each the mirror image of an L amino acid; glycine, which has no stereocentre, has none.

Groups bonded to residues #

A ligand bonded to a residue is a glycan when it is a sugar, a lipid or a cofactor when its component is listed here, an additive adduct when it is an additive, and a covalent ligand otherwise.

A sugar's category comes from the residue and atom it is bonded to, the first matching rule:

Ionizable groups of the pH analysis #

The charge is that of the charged form (an acid's when it has lost its proton, a base's when it holds one); the desolvation shift is the one of a fully buried group.

Residue names whose side chain titrates, with the protonation-state names of simulation files (a HIP holding a phosphorus atom is phosphohistidine and does not titrate):

A chain that starts with ACE, FME, FOR, PCA has no free N-terminal amine; one that ends with NH2 or NME has no free C-terminal carboxyl.

The pH analysis takes the metal elements of the ligand classes above (in lone ions, clusters and cofactors alike: a group within 2.8 Å of one is coordinated). A lone metal ion carries the charge its component's formula states or, when the formula states none, the charge of the element's usual simple cation in water; elements whose usual forms in water are oxo ions or several get none:

Environments #

Membrane slab classes #

The membrane placement weighs each residue's exposed surface by its class:

Through the API #

curl -s 'https://YOUR-DOMAIN/api/bio/nomenclature'

GET /api/bio/nomenclature answers with the identifiers and every table on this page, as JSON, cached for an hour:

The route needs no database and takes no parameters. Through MCP it is reached with api_request (path: /api/bio/nomenclature); list_api_routes lists it under Names and identifiers.

Limits #

• The lists are short on purpose. They hold the components that matter most often, not the CCD's tens of thousands. A buffer missing from the additive list is a ligand; a saccharide missing from the sugar list is a sugar only when its file types it as one (mmCIF does, PDB files do not); a modified residue missing from the modified-residue table is nonstandard.
• Parents are read from the table, not the file. Files can name a modified residue's parent (MODRES in PDB files, _pdbx_struct_mod_residue in mmCIF), but GL1X reads parents only from its amino-acid and nucleotide tables, the same for every format; a modified residue missing from them is written X or N even when its file names the parent.
• Classes are a convenience, not chemistry. They describe a component's usual role: citrate is listed as an additive and is also a substrate of citrate synthase; an iron-sulfur cluster is a ligand because it is not a single metal atom.
• Names are as good as the files. GL1X reproduces a file's names, including old or unusual ones, and a component a PDB file uses without a HETNAM record has no name in GL1X.

References #

• Berman HM, Westbrook J, Feng Z, et al. (2000) The Protein Data Bank. Nucleic Acids Res 28:235–242.
• Creasy DM, Cottrell JS (2004) Unimod: Protein modifications for mass spectrometry. Proteomics 4:1534–1536.
• IUPAC-IUB Joint Commission on Biochemical Nomenclature (1984) Nomenclature and symbolism for amino acids and peptides. Recommendations 1983. Eur J Biochem 138:9–37.
• Kabsch W, Sander C (1983) Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22:2577–2637.
• Nomenclature Committee of the International Union of Biochemistry (1985) Nomenclature for incompletely specified bases in nucleic acid sequences. Recommendations 1984. Eur J Biochem 150:1–5.
• Nomenclature Committee of the International Union of Biochemistry and Molecular Biology (1992) Enzyme Nomenclature 1992. Academic Press, San Diego.
• Prohaska T, Irrgeher J, Benefield J, et al. (2022) Standard atomic weights of the elements 2021 (IUPAC Technical Report). Pure Appl Chem 94:573–600.
• The UniProt Consortium (2023) UniProt: the Universal Protein Knowledgebase in 2023. Nucleic Acids Res 51:D523–D531.
• Thurlkill RL, Grimsley GR, Scholtz JM, Pace CN (2006) pK values of the ionizable groups of proteins. Protein Sci 15:1214–1218.
• Varadi M, Anyango S, Deshpande M, et al. (2022) AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res 50:D439–D444.
• Wang M, Audi G, Kondev FG, Huang WJ, Naimi S, Xu X (2017) The AME2016 atomic mass evaluation (II). Tables, graphs and references. Chinese Phys C 41:030003.
• Westbrook JD, Shao C, Feng Z, Zhuravleva M, Velankar S, Young J (2015) The chemical component dictionary: complete descriptions of constituent molecules in experimentally determined 3D macromolecules in the Protein Data Bank. Bioinformatics 31:1274–1278.
• wwPDB consortium (2019) Protein Data Bank: the single global archive for 3D macromolecular structure data. Nucleic Acids Res 47:D520–D528.