1. Introduction to UniRule format

UniRule is a format describing conditional annotation templates (rules) used by UniProtKB/Swiss-Prot automated annotation projects. It defines annotation which will be generated for (selected) predicted features.

Structure

Each UniRule entry consists of the following parts.

• The Header section contains the UniRule accession number(s), the data class, identifier(s) of the feature(s) that trigger(s) the rule, as well as some basic information on the rule.
• The Annotation section. All UniProtKB/Swiss-Prot annotation relevant for a UniRule is indicated. Not all annotation lines apply to all true hits. Therefore, case statements restrict all or part of the annotation to further conditions like the taxonomic group or the size of a protein.
• The Computing section covers information on a protein, domain or site relevant to programs. It is supposed to be checked automatically and it is still incomplete.
• A // (terminator) line which contains no data or comments and designates the end of an entry.

Application

‘Protein rules’

Used to annotate protein family : HAMAP rules.

‘Domain/Site rules’

Used to annotate protein domain or site : ProRules.

Explanation of terms

‘motif’

Any sequence feature predictor or discriminator, such as a pattern, profile or profile-HMM.

‘taxcode’

A mnemonic code of at most 5 alphanumeric characters specific for a taxonomic species used in UniProtKB/Swiss-Prot.

2. Header section

The Header section contains technical information about the UniRules.

2.1 AC line

Each UniRule starts with one or more AC (ACcession number) lines. The format of the accession number is for intance ‘PRU’ followed by 5 digits for ProRules and ‘MF_’ followed by 5 digits for HAMAP rules.

Protein rule: mandatory

Domain/Site rule: mandatory

AC   primary accession number;[ secondary accession numbers;]

AC   PRU00001;

AC   PRU00002; PRU99998; PRU99999;

AC   MF_00022;

Note: The UniRule file name in the CVS directory corresponds to the primary accession number of a UniRule.

2.2 DC line

The DC (Data Class) line specifies which type of data the rule annotates.

Protein rule: mandatory

Domain/Site rule: mandatory

DC   data_class;[ auto]

DC   Protein; auto

DC   Domain;

DC   Site;

The data_class is either of the following values:

• ‘Protein’
• ‘Topo_domain’
• ‘Domain’
• ‘Site’

‘Protein’

Refers to a rule that enables complete annotation of a UniProtKB/Swiss-Prot entry.

‘Topo_domain’

The rule applies to a domain of topological meaning, like an extracellular region. A ‘Topo_domain’ may contain any ‘Domain’ within its range.

‘Domain’

Refers to a rule that enables domain annotation of a UniProtKB/Swiss-Prot entry.

Domains are generally not allowed to overlap. However in rare cases it is possible that a domain is located within another domain; e.g. EH domain and EF-hand. In that situation, the annotation of the smaller domain must be triggered in the rule for the longer domain.

‘Site’

Refers to a rule that enables site annotation of a UniProtKB/Swiss-Prot entry. One or more sites may be located within a domain. If a site within a domain is conserved, it is usually triggered in the domain rule (See section DR line).

A Protein rule has precedence over a Domain/Site rule.
The optional trailing term ‘auto’ indicates that a UniRule can be applied to automated annotation. If absent, it means that the UniRule should be used only for the support of manual annotation, which is the case when a method has a significant number of false positives or the output needs further manual completion.

2.3 TR line

The TR (Trigger) line describes which (selected) sequence analysis features trigger the application of the curent UniRule. Each UniRule may contain one or more TR lines. Each feature name should appear only once in the TR line in the whole UniRule database, as one type of feature is expected to trigger only one rule.

Protein rule: mandatory

Domain/Site rule: mandatory

There are 2 types of TR lines:

2.3.1 TR motif line

TR   dbname; identifier1; identifier2; nbhits; level=level

TR   PROSITE; PS50075; ACP_DOMAIN; 0-1; level=0

TR   HAMAP; MF_00401; -; 1; level=0

TR   General; Transmembrane; -; 1; level=0

TR   IPRO; IPR009003; -; 1; level=0

• The dbname is generally the name of a motif database, for example ‘PROSITE’ or ‘HAMAP’. ‘General’ is used for rules which refer to a specific feature type, independently of the prediction method used.
• The identifier1 is the feature name e.g. the unique identifier of the motif in the given database, which is generally an accession number. For ‘General’ methods a term indicates the feature identifier which triggers the rule, for example Transmembrane, Signal. If dbname is ‘IPRO’, the identifier1 is an InterPro number, and all motifs under that InterPro number will be triggers.
• The identifier2 is a secondary identifier for the motif, which is generally an entry name, and which is empty (‘-’) in trigger lines of type ‘HAMAP’ and ‘General’.
• The nbhits field indicates the expected number of hits to the motif. Obviously, the rule is only triggered if at least one hit is found. However, since motifs referred to in the TR line should not be repeated in the DR line, this field controls whether a FALSE_NEG cross-reference should be created if another TR line triggers the rule but no hit is found to the motif in the given TR line (See section DR line for a discussion). Legal values are ‘1’ and ‘0-1’, and the latter value is only valid if there are multiple TR lines.
• The level indicates the minimum level that a hit must have to trigger the rule. Note: obsolete, not used anymore!

2.3.2 TR Metamotif line

TR   Metamotif; mmsearch-options; metamotif

TR   Metamotif; -; Signal>=GPI_anchor

Technical note: Now metamotifs are detected (through mmsearch) by anabelle psat; they are just ’classical’ gff psat features. Therefore metamotifs detection is not tied anymore to the rule evaluation step. The metamotif TR should not be included in a condition to test if its required features are present, this is now useless/obsolete (as the metamotif was detected by psat, it implies that its features were present).

The metamotif field should correspond to the name of a detectable metamotif feature. The metamotif feature name is at the same time the metamotif description. It describes the arrangements of sub-features (by their name) in a sequence. The ’available’ metamotifs are listed in $ANABELLE/data/metamotifs.dat This file is used by mmsearch, and is created at release (anabelle_update.sh) by extracting TR metamotif lines from CVSed UniRules.

The field mmsearch-options used to control the behavior of the metamotif search automaton (during ’picohamap’ times?). It is now obsolete, just use ’-’.

2.4 XX line

This separator line conveys no meaning. It is used to separate blocks of lines.

2.5 Names line

The Names line indicates the name(s) of the motif described by the rule. The first name is the one used in UniProtKB/Swiss-Prot, synonyms are listed below, one name per line. If there is no specific name for a motif, the term ‘Undefined’ should be inserted as a placeholder.

Protein rule: 1 name is mandatory

Domain/Site rule: mandatory

Names: Undefined

Names: Synonym1

 Synonym2

 …

Names: Nacht domain

 NACHT-NTPase domain

 Nucleoside triphosphatase domain

2.6 Function line

The Function line indicates the ‘generic’ function of a domain. Usually it should not be longer than one line, but in case it is, the lines following the first one will be indented by a space.

Protein rule: forbidden

Domain/Site rule: mandatory

Function: Undefined/Unknown

Function: text

Function: Protein-binding

Function: DNA-binding

Function: Inhibits fibrinogen interaction with platelet receptors in

 snake venoms

2.7 Internal comments

Comments concerning the rule, which are for internal use, and do not appear in published versions. It should appear on lines prefixed with two asterisks and a space. They may appear at the very end of the Header section, and throughout the Annotation section and the Computing section. However, it is good practice to put such comments at the end of the Header section where they are immediately visible, and only put them exceptionally in the Annotation and Computing sections.
Example:

** SALRD unusually long in the C-terminus, flagged as atypical.

3. Annotation section

The Annotation section includes all UniProtKB/Swiss-Prot annotation lines that can be applied to a rule match. Additionally there can be lines which indicate condition statements (‘case’, ‘else case’, ‘else’, ‘c?’, ‘c!’) and a line which indicates the motif or alignments, according to which the feature positions are calculated (See section FT From line). The line order is the same as in UniProtKB/Swiss-Prot.

3.1 ID line

The ID line contains the mnemonic code for a protein name used in the entry name of a UniProtKB/Swiss-Prot entry.

Protein rule: mandatory

Domain/Site rule: forbidden

ID   protein_name_code

ID   ACKA

Note: If specified code is ’Ynnn’ it will be replaced by ’Y’ followed by the OLN (OrderedLocusName) number of the annotated protein.

3.2 DE line

The DE line corresponds to the description line of a UniProtKB/Swiss-Prot entry. To define DE annotation that should be added to existing data (from original entry and/or from previous rule element applications) instead of replacing it: prepend DE block with ‘+’.

Protein: mandatory

Domain/Site: optional

DE   Description.



DE   + partial_description

DE   RecName: Full=Putative 3-methyladenine DNA glycosylase;

DE            EC=3.2.2.-;



DE   + RecName: EC=2.7.3.-;

The DE line may contain the placeholder ‘<locus_tag>’, which is to be replaced (in annotation output) by the Ordered Locus Name (OLN) of the entry (or by its ORFNames if no OLN was found; if nothing can be found: will be left as it is but a warning will be generated).
It may also contain the placeholder ‘<gene_name>’ which is to be replaced by the entry gene name (GN Name).

Example:

DE   RecName: Full=Hemerythrin-like protein <locus_tag>.

The DE line may contain the @gn(ANYGENENAME) placeholder/command, which is to be replaced by the correct casing and gene count numbering (according to the entry taxonomy) of the specified gene name.

3.3 GN line

The GN line contains the common gene name (and optionally synonyms) of a protein family, when one exists.

Protein: optional

Domain/Site: forbidden

GN   Name=name;[ Synonyms=synonym[, synonym]…;]

GN   Name=acpD;

GN   Name=groS; Synonyms=groES;

3.4 CC line

The CC line(s) contains all applicable comment lines of a UniProtKB/Swiss-Prot entry.

Protein: optional

Domain/Site: optional

CC   -!- topic: text.

CC   -!- SIMILARITY: Belongs to the ABC transporter family.

The CC line topic ‘SIMILARITY: Contains’ is often not complete. The hash sign ‘#’ in the text stands for the number of hits found for the given motif and is replaced during the automated annotation procedure; the term ‘repeat’, ‘domain’ or ‘zinc finger’ will be set to plural if more than 1 true positive hit was found of them in a protein.

The CC line may contain :

• The placeholder ‘<gene_name>’, which is to be replaced (in annotation output) by the entry gene name (GN Name).
• The #(ftdesc,text) placeholder e.g. #(manganese,ion) which is to be replaced by the number of distinct instances of the specified ft description element (from FT lines takes the highest number corresponding to the specified element; works only for numered ft desc (with ‘#’); (See section FT line) plus if specified, the ‘text’ pluralized if more than 1 instances were found. e.g. 2 manganese ions. If the text (and the ,) is ommited, the ftdesc itself will be pluralized if needed.
• The @gn(ANYGENENAME) placeholder/command, which is to be prelaced by the correct casing and gene count numbering (according to the entry taxonomy) of the specified gene name.
• The #{[Aaa-]POS[:template_name]} placeholder will be replaced by the position mapped - from the position POS in the match region or, in protein rule mode, the defined entry template - to the correct numbering in the target sequence, prefixed with the matched amino acid (3 letters code). If Aaa amino-acid is specified (3 letters code), a warning will be generated if the mapped residue does not correspond to the specified one. If a template_name is defined, the corresponding sequence template will be used for the mapping (needed for disambiguation when there are several entry templates! e.g. for a HAMAP alignment with several defined templates)

CC   -!- PTM: The phosphorylation of #{120} activates ...

CC   -!- PTM: Phosphorylation at #{Ser-187:FEN1_HUMAN} by CDK2 ...

CC   -!- PTM: The reversible ADP-ribosylation of #{Arg-101} inactivates ...

The latter becomes for example:

CC   -!- PTM: The reversible ADP-ribosylation of Arg-112 inactivates ...

...

3.5 DR line

The DR line main usage is to trigger ’child’ rules in order to avoid duplication of rule content. It has not much to do with ’real’ DR UniProtKB/Swiss-Prot annotation lines (in Anabelle, DR annotation is only performed for PROSITE motifs). The format of the DR line is similar to that of a TR line (See section TR line).

Protein rule: optional

Domain/Site rule: optional

DR   type/dbname; feature name; identifier; nbhits; trigger=[yes|strict|no]

DR   PROSITE; PS00419; PHOTOSYSTEM_I_PSAAB; 1; trigger=no

DR   PROSITE; PS00010; ASX_HYDROXYL; 0-1; trigger=no

DR   General; Signal; -; 0-1; trigger=strict

DR   General; Transmembrane; -; 10-11; trigger=yes

• The type/dbname field indicates the database name - for example ’PROSITE’ or ’HAMAP’ - or ’General’ (for non database features e.g. Signal) or ADD_TOPO_DOMAIN (see below).
• The feature name field indicates the name of the feature that should be used (e.g. Transmembrane, motif accession number, etc...) to propagate annotation through the triggering of its associated rule.
• The identifier is a secondary identifier (ID) for the motif which is empty (‘-’) in trigger lines of type ‘HAMAP’ and ‘General’ (for information purposes, not checked/used).
• The nbhits field indicates whether the "child feature" is mandatory, and the number of (’hits’) features which are expected. The value of this field may be a number (‘1’), or a range (‘0-1’). A range in the form ‘number-unlimited’ indicates that the number of matches is unlimited.
  If the number of expected features does not match the number of detected ones a warning (**HW SAM line) will appear in the generated annotation.
  If a PROSITE cross-reference is found in a mandatory (i.e. where nbhits does not contain zero) TR or DR line but no match is found, a DR line will be created which includes the status qualifier ‘FALSE_NEG’. In rules of the data class ‘Protein’ it is further possible to include cross-reference lines with the status qualifier ‘FALSE_POS’ for matches to which a rule does not refer.
  If the nbhits field of the DR line is ‘0-1’, the feature applies to the domain only in some proteins (it won’t ’complain’ if none is found). In that case for PROSITE motifs, DR FALSE_NEG will not be created if no match is found.
• The trigger, must be set to either of ‘yes’, ‘strict’ or ‘no’. The values ‘yes’ or ‘strict’ indicates that the rule associated with the specified features should be triggered to generate annotation. With ’strict’, only selected features (by the analysis result automatic selection module) will be used. With ’yes’, selected features will also be used preferentially, but if none is found, all (non-selected) features of the specified type will be used!
  The value ‘no’ indicates that features (overlapping, if parent rule is not a protein type rule, all, if parent rule is of protein type) with the specified feature name will not be used to generate annotation.

Notes:

• To specify TOPO_DOMs (by triggering topological domain associated rules) use DR ADD_TOPO_DOMAIN
  Examples:

  DR ADD_TOPO_DOMAIN; Cytoplasmic; -; 2; trigger=yes DR ADD_TOPO_DOMAIN; Periplasmic; -; 1; trigger=yes

  The first line will set the Nter TOPO_DOM to ’Cytoplasmic’, the second line specifies the alternative compartment ’Periplasmic’ to propagate alternate Cytoplasmic / Peroplasmic TOPO_DOMs between (selected) Transmembranes. N.B. the ’fake’ ADD_TOPO_DOMAIN type/dbname is a flag for the sequence analysis module (that peaks inside the rule) to tell it to create corresponding topological domain sequence analysis "results" between selected transmembrane.

• DR lines can be used to trigger rules associated with any kind of feature, not just motif/profile matches (see examples). For example it can be used to trigger Signal, Transmembrane , topological domains (see above) etc...
• DR lines can be inside cases...
• the used features will not generate annotation by themself (hence a second time) after they are ’used’ through those rule DR lines.
• A protein type rule will prevent all lower (e.g. domain) rules from independently being triggered by their associated feature. Therefore "under" a protein rule: the only non protein rules that will be used are the ones associated to features specified in those DR (with trigger=yes|strict) lines.
• The DR line could also be used, with trigger=no, to prevent the generation of specific child annotation. It will generate warnings if the number of expected features does not match the number of detected ones. If the feature name is a PROSITE motif: It will also set the status of DR PROSITE lines (see above "The nbhits field").
• Through a protein type rule, the DR directive will consider could rules associated with any feature detected within the same protein. Through a non protein type rule, only the features overlapping with the ’current’ rule associated features will be considered.
• Considered features (the ones used to trigger annotation) will prevent the subsequent triggering / usage of any other overlapping features specified in DR lines, except for the ADD_TOPO_DOMAIN. Features are examined following the order in the DR lines. In the first example, the Signal (if present) will be used first and so prevent the usage of any overlapping transmembrane. Transmembranes that do not overlap with the used signal will then be used after.
  For topological domains (via ADD_TOPO_DOMAIN): place them first in the DRs (so that they are not blocked by a previously triggered overlapping feature), they won’t block the triggering of other features.

3.6 KW line

The KW line contains all applicable keywords for a UniProtKB/Swiss-Prot entry, one per line.

Protein rule: optional

Domain/Site rule: optional

KW   keyword

…

KW   Transferase

KW   Kinase

3.7 GO line

The GO line contains all applicable Gene Ontology terms, one per line.

Protein rule: optional

Domain/Site rule: optional

GO   accession-number; aspect:term

…

GO   GO:0019104; F:DNA N-glycosylase activity

GO   GO:0006281; P:DNA repair

3.8 FT line

The FT (Feature table) line contains applicable features for a UniProtKB/Swiss-Prot entry. The feature positions are calculated by the automated annotation program based on the rule and the motif match positions.

Protein rule: optional

Domain/Site rule: optional

FT   From: template-id (template-accnumber) OR unique identifier for a motif

FT   key      from     to       desc.

[FT   [Optional;] [Group: n;] [Condition: pattern]]

FT   From: CARB_ECOLI (P00968)

FT   DOMAIN     Nter    403       Carboxyphosphate synthetic domain.



FT   From: PS00385

FT   SITE          6      6       By similarity.

3.8.1 FT From line

The FT From line is mandatory above the FT feature line. It defines the "template" / "frame of reference" to allow the rule specified coordinates to be mapped into the target sequence coordinates, for all subsequent FT feature lines. The template must be one of the following: ’any’, the unique identifier for a motif, a protein identifier, or a metamotif.

FT   From: unique identifier for a motif [:regexp]

FT   From: entry_name (accession_number) [:regexp]

FT   From: metamotif [:regexp]

FT   From: any [:regexp]

FT   From: PS50234

FT   From: ACP_ECOLI (P02901)

FT   From: PS50021=7,91=PS50021

FT   From: any

FT   From: PS50217 :([KR])[^LI].*([KR])[^LI]([LI]).*([LI])

If (accession_number) is given = the template/frame of reference is a UniProt protein = the numbering is based on this protein, the subsequent positions will be mapped to the annotated target sequence.
If (accession_number) is not given = the frame of reference is the trigger (motif) match region. If a specific motif identifier is given the subsequent FTs will only be propagated if the current trigger is of the specified kind (this is useful when one rule as multiple triggers...). If "any" is used, no filtering by trigger name will be performed.
The optional regexp, can be used in conjonction with $1-9 position "keys" in subsequent FTs to annotate floating postions.

N.B. If the template is a metamotif, it must be present verbatim in the TR line.

3.8.2 FT feature line

The FT feature line defines the actual FT lines that are to be propagated in member entries.

FT   key      from     to       desc.

FT   CHAIN      to+1   Cter       <name>.

FT   LIPID         1      1       GPI-anchor amidated <residue_name>.

FT   DOMAIN     from     to       Laminin G-like #.

FT   TOPO_DOM   Nter      6       Periplasmic (Potential).

FT   DOMAIN        ?    8+1       EGF-like #.

FT   From: PS50217 :([KR])[^LI].*([KR])[^LI]([LI]).*([LI])

FT   REGION       $1      $2       Basic motif #.

FT   REGION       $3      $4       Leucine-zipper #.

Placeholders (in the description)

• The ‘<name>’ placeholder will be replaced (in the annotation output) by the protein RecName (Full) (DE line 1st elem).
• The ‘<gene_name>’ placeholder will be replaced by the entry gene name (GN Name).
• The ‘@gn(ANYGENENAME)’ placeholder/command, will be replaced by the correct casing and gene count numbering (according to the entry taxonomy) of the specified gene name.

  FT REGION 101 124 Necessary for interaction with @gn(ABC-1).

• The ‘<residue_name>’ placeholder will be replaced by the chemical name of the sequence residue on which the feature resides. Thus the following line in a UniRule:

  FT LIPID 1 1 GPI-anchor amidated <residue_name>.

  may be transformed by samann.pl into the following UniProtKB/Swiss-Prot line:

  FT LIPID 300 300 GPI-anchor amidated aspartate.

• The #{[Aaa-]POS} placeholder will be replaced by the position mapped - from the position POS in the match region or, in protein rule mode, the defined current entry template (c.f. "FT From") - to the correct numbering in the annotated sequence, prefixed with the matched amino acid (one letter code). If an Aaa amino-acid is specified (3 letters code), a warning will be generated if the mapped residue does not correspond to the specified one.

  FT CROSSLNK 238 264 Tryptophyl-tyrosyl-methioninium (Tyr-Met) FT (with #{Trp-90}) (By similarity).

  The latter will generate for example:

  FT CROSSLNK 240 266 Tryptophyl-tyrosyl-methioninium (Tyr-Met) FT (with W-92) (By similarity).

• for FT ACT_SITE description: text between square brackets [ ] will be shown only for entry that have more than one enzymatic activity (based on the number of ‘(EC num’ fields found in the entry DE line). If the entry has 1 or no defined enzymatic activity, the whole text (and the brackets) won’t be annotated.

  FT ACT_SITE 6 6 [For protease activity] By similarity.

  may be transformed by samann.pl into the following UniProtKB/Swiss-Prot line: If the entry has several EC:

  FT ACT_SITE 506 506 For protease activity (By similarity).

  Note that if the text is added, the qualifier (By similarity), is put between brackets (as it is not alone).
  If the entry has only 1 enzymatic activity:

  FT ACT_SITE 506 506 By similarity.

• The ‘#’ placeholder may be used to assign automatic numbers in the desc field. If the feature occurs over once with the same text before the occurrence of the placeholder, the placeholder is replaced with subsequent numbers.
• The ‘#n’ placeholder, where n is a number starting from 1, should be used when it is needed to reference distinct ligands with the same name. For example, the following made-up rule:

  FT DOMAIN from to Foobar #. FT METAL 87 87 Iron #1 (By similarity). FT METAL 118 118 Iron #1 (By similarity). FT METAL 118 118 Iron #2 (By similarity). FT METAL 180 180 Iron #2 (By similarity).

  Would yield the following annotation in a protein with a single match:

  FT DOMAIN 1 200 Foobar. FT METAL 87 87 Iron 1 (By similarity). FT METAL 118 118 Iron 1 (By similarity). FT METAL 118 118 Iron 2 (By similarity). FT METAL 180 180 Iron 2 (By similarity).

  And the following annotation in a protein with two matches:

  FT DOMAIN 1 200 Foobar 1. FT DOMAIN 201 400 Foobar 2. FT METAL 87 87 Iron 1 (By similarity). FT METAL 118 118 Iron 1 (By similarity). FT METAL 118 118 Iron 2 (By similarity). FT METAL 180 180 Iron 2 (By similarity). FT METAL 287 287 Iron 3 (By similarity). FT METAL 318 318 Iron 3 (By similarity). FT METAL 318 318 Iron 4 (By similarity). FT METAL 380 380 Iron 4 (By similarity).

3.8.3 FT constraints line

The FT constraints line (also known as the FT Condition line) gives constraints on the FT line immediately above it.

FT   [Tag: tagname[, tagname]…;] [Optional;] [Group: n;] [Condition: pattern]

FT   BINDING      37     37       Phosphopantetheine (By similarity).

FT   Tag: phospho; Condition: S

The ‘pattern’ is specified in the PROSITE pattern format with the addition

that the character ‘*’ may be used to specify an unconstrained range, e.g. ‘C-x*-C’. The region of the sequence corresponding to the feature must exactly match this pattern.

For the consistency of annotation, multiple FT lines that should be applied either all together or not at all should be grouped within a ‘Group’, to constrain the common presence of all sites. This group can the be referenced by case statements, for instance in the relevant KW and CC lines that depend on the presence of the feature.

Example:

case <FTGroup:1>

KW   GTP-binding

end case

XX

case <OC:Bacteria>

FT   From: IF2_ECOLI (P02995)

FT   DOMAIN      392    540       G-domain.

FT   Group: 1

FT   NP_BIND     398    405       GTP (By similarity).

FT   Group: 1; Condition: G-H-V-D-H-G-K-T

FT   NP_BIND     444    448       GTP (By similarity).

FT   Group: 1; Condition: D-T-P-G-H

FT   NP_BIND     498    501       GTP (By similarity).

FT   Group: 1; Condition: N-K-[LIVCM]-D

end case

Note: One FT line can be part of several FTGroups. If at least one of those groups is complete, the FT line passes its FTGroup constraint (implicit OR).

Example:

FT   DISULFID     25     31       By similarity.

FT   Group: 1; Group: 2; Condition: C-x*-C

The ‘Optional’ label can be used to indicate that the absence of a feature should not be considered a trigger for warnings in annotation programs. It can only be present if a ‘Condition’ pattern is supplied.

Example:

FT   BINDING      37     37       Phosphopantetheine (By similarity).

FT   Optional; Condition: S

A ‘Tag’ name (or several) can be given to the feature (N.B. distinct features can have the same tag name) for easier use of case against specific features (with case <FTTag:tagname>) See section Case statement.

Example:

FT   DISULFID     48     51       Redox-active (By similarity).

FT   Tag: disulf, redox; Condition: C-x*-C

4. Computing section

The Computing section contrasts with the Annotation section in that the line identifiers are no longer restricted to 2 letters. The information starts in the line of the line identifier, the following lines are indented by 1 space.

Line identifier1: info in line1

 following lines are indented

Line identifier2: info in line1

 following lines are indented

 …

Not all line types are relevant to any data class (See section DC line).

4.1 Warn line

Specify a warning that should be generated when the rule is used for automatic annotation. Most often used within a case statement to indicate the occurrence of an inconsistency that cannot be solved by the rule, or that some annotation should be completed manually by a curator. The SAM module transfers the text of Warn lines into the ‘**HW’ section of a UniProtKB/Swiss-Prot entry.

Protein rule: optional

Domain/Site rule: optional

Warn: text

case <OC:Proteobacteria>

Warn: Check manually domain bounds

end case

4.2 Chop line

Range by which the bounds of a domain may be chopped in order to annotate successive domains in an exactly consecutive manner. This line can only be used by programs if the complete size of the domain can be annotated; generally it will not be possible to use it with a pattern that covers only part of the domain.

Protein rule: forbidden

Domain/Site rule: mandatory

Chop: Nter=max; Cter=max;[ Xter(motif)=max;]*

Chop: Nter=0; Cter=3;

Chop: Nter=1; Cter=unlimited;

Chop: Nter=0; Cter=0; Nter(Signal)=50;

• max indicates the maximum number of positions, by which the N- terminal or C-terminal may be trimmed to be able to annotate adjacent identical (by name) domains which overlap slightly. max can be either of 0 (by default), a positive value, or the word ‘unlimited’.
• ‘Nter’ and ‘Cter’ indicate the number of positions that may be trimmed when the triggering motif is found adjacent to any other identical (by name) motif. Additionally, specific adjacent motifs for which trimming is also allowed can be indicated by giving the motif name in braces, for example ‘Nter(Signal)’.

4.3 Size line

The Size line indicates the size relevant to a protein family or motif. For entries of the data class ‘Protein’, the minimal and maximal size of proteins matching the rule are listed. For entries of the data class ‘Domain’, this line contains the size range of the complete domains annotated in UniProtKB/Swiss-Prot. Members that are strongly divergent in size may be excluded from the range. A size may be specified as ‘unlimited’.

Protein rule: mandatory

Domain/Site rule: mandatory

Size: minimal_size-maximal_size;

Size: fixed_size;

Size: 176-239;

Size: 13-unlimited;

Size: unlimited;

4.4 Related line

Lists UniRules that are known to be similar in sequence, and risk to produce cross-matches. If the string ‘!’ or ‘!!’ is appended to the name of a rule, it means that the rule listed in the Related line supersedes the current rule, i.e. matches to the current rule should be disregarded if a match with the listed rule is found: ‘!’ in an overlapping region; ‘!!’ anywhere on the protein.

The marker ‘!’ is particularly useful when two different rules exist for a ‘short’ and a ‘long’ version of the same protein (as occurs sometimes in HAMAP families). ‘Long’ proteins will match both profiles; under these circumstances the ‘longer’ UniRule should contain the ‘!’ marker to supersede the shorter UniRule.

Protein rule: mandatory

Domain/Site rule: mandatory

Related: None;

Related: Protein[!][!];[ Protein[!][!];]…

Related: MF_00492; MF_00493; MF_00494;

Related: MF_00344!;

Related: ANA00003!!;

4.5 Repeats line

The observed number of repetitions of a domain or site in a UniProtKB/Swiss-Prot entry. The number may be specified as ‘unlimited’.

Protein rule: forbidden

Domain/Site rule: mandatory

Repeats: value;[ no keyword;]

Repeats: min-max;

Repeats: 1;

Repeats: 2-4;

Repeats: unlimited; no keyword;

4.6 Topology line

Specifies the subcellular location(s) in which a domain or site may occur.

Protein rule: optional

Domain/Site rule: mandatory

Topology: Undefined;

Topology: location;

Topology: Not cytoplasmic;

4.7 Template line

Lists accession number(s) of characterized proteins which were used to build the UniRule (Note: indicative only). Uncharacterized protein families do not necessarily have a template, this is noted as ‘Template: None;’. Note that in many cases the propagated annotation is a subset of that found in the characterized entries.

Protein rule: mandatory

Domain/Site rule: forbidden

Template: accession_number;[ accession number;]…

Template: None;

Template: Undefined;

Template: P12345;

Template: None;

Template: Undefined;

4.8 Example line

One or more example entry targeted by the rule.

Protein rule: forbidden

Domain/Site rule: mandatory

Example: accession_number;[ accession number;]…

Example: Undefined;

Example: P12345;

Example: Undefined;

4.9 Scope block

Lists the taxonomic classes in which a rule match may be found.

Protein rule: mandatory

Domain/Site rule: mandatory

Scope:

 kingdom[; sub-taxon]

  [except sub-taxon

  …]

  [not in taxcode[, taxcode]…]

 …

Scope:

 Bacteria; Proteobacteria

  except Enterobacteriales

  except Pasteurellales

 Bacteria; Actinobacteria

 Archaea

  not in ARCFU, HALN1, METTH, METJA, PYRAB, PYRHO, SULSO, SULTO,

  THEAC, THEVO

 Plastid

The taxonomic classification is composed of the kingdom, optionally followed by the name of a sub-taxon, to further limit the application of the UniRule to any taxonomic level. Valid values for kingdom are: ‘Eukaryota’, ‘Bacteria’, ‘Archaea’, ‘Viruses’, ‘Bacteriophage’, ‘Plastid’ and ‘Mitochondrion’. The two latter values designate proteins encoded in the organelle genome but not proteins encoded in the nucleus and targeted to the organelle.

If it has been assessed with certainty that a UniRule is not represented in:

• A taxonomic group, its name may be indicated in the ‘except’ field.
• A complete proteome, its taxcode may be indicated in the ‘not in’ field.

Note: Plasmid is not defined as kingdom; there is a separate line type (See section Plasmid line).

4.10 Fusion block

Lists UniRules to which a given UniRule may be fused in some instances.

Protein rule: mandatory

Domain/Site rule: forbidden

Fusion:

 NT: None

 CT: None

Fusion:

 NT: Protein[; Protein]…

 CT: Protein[; Protein]…

Fusion:

 NT: None

 CT: MF_00222 (aroE); <Unknown>

4.11 Duplicate line

Lists the organisms for which the motif which triggers the rule is found in multiple copies.

Protein rule: mandatory

Domain/Site rule: forbidden

Duplicate: None

Duplicate: in taxcode[, taxcode]…

Duplicate: in ANASP, CAUCR, LACLA, RHILO, RHIME, STAAU, SYNY3

4.12 Plasmid line

Lists the organisms for which the motif which triggers the rule is found encoded on a plasmid.

Protein rule: mandatory

Domain/Site rule: forbidden

Plasmid: None

Plasmid: in taxcode[, taxcode]…

Plasmid: in RHIME

4.13 Comments block

Comments concerning the rule, which should be made visible to the public.

Protein rule: mandatory

Domain/Site rule: mandatory

Comments: None

Comments: comment_text

Comments: NUDIX-like C-terminal domain in SYNY3

5. Control statements

5.1 Case statement

Format:

case <condition>[ and|or [not] [defined] <condition>]…

else case <condition>[ and|or [not] [defined] <condition>]…

else

end case

The ‘case’ and ‘else case’ lines include conditions that must be met for the lines below it to be applied, until the next ‘else case’, ‘else’ or ‘end case’ statement. Condition lines (c! and c?, see below) do not break the latest case statement.

Note: It is not possible to use a ‘case’ statement within a ‘case’ statement, but it is possible to use the condition lines c! or c?.

Types of cases:

• OS/OC/OG: on taxonomy and organelle (OS, OC and OG lines):

  case <OG:Chloroplast> or <OC:Cyanobacteria> case not <OG:Chloroplast> and not <OG:Cyanelle> case <OC:Archaea> case <OC:Bacteria> case <OS:Staphylococcus aureus>

  Note for conditions on organism names (‘case <OS:taxon>’): the organism name matches also subspecies, i.e. organisms with the same name followed by a space and then any text. For example, ‘Staphylococcus aureus’ matches ‘Staphylococcus aureus RF122’, but ‘Salmonella typhi’ does not match ‘Salmonella typhimurium’.

• FT: on a feature (FT block) propagated from the rule (a pattern that must be matched by the feature can also be specified):

  case <FT:1> case <FT:1=A-x-x>

  Note: In the FT lines, this condition can only be used if the targeted FT is itself not in a FT[Group] case!

  Note: The targeted feature number corresponds to its position/order in the FT block. Caution: FT lines in a FT[Group] case are also numbered according to their relative position in the FT lines, but the numbering starts at 1 + the number of FT element - not in a FT[Group] case (it is therefore simpler to put those FT elements at the bottom of FT lines)...

• FTGroup: on the fact that a feature group was propagated: (<FTGroup:n> See section FT constraints line)

  case <FTGroup:1> → true if all features in Group 1 were be propagated

  Note: In the FT lines, this condition can only be used if the targeted FTGroup is itself not in a FT[Group] case!

• FTTag: on the fact that a feature with a certain tag was propagated: (Tag: tagname. See section FT constraints line)

  case <FTTag:phospho> → true if at least one feature with the ‘phosho’ tag was propagated

• (Any)Feature: on the presence of a feature in the GFF file:

  case <Feature:PS50084> case <AnyFeature:PS50084> case <Feature:Transmembrane>2> → feature must be present over two times

  The operators ‘>’, ‘<’, ‘==’, ‘>=’, ‘<=’ are supported.

  The difference between ‘Feature’ and ‘AnyFeature’ is as follows. ‘Feature’ only refers to the (rule) triggering feature + features that overlap (by at least 50%) with it. ‘AnyFeature’ refers to all features matching the sequence.

  Those conditions refer to both selected and unselected features, except if an operator (>,<,==,>=,<=) is present: here only selected features will be examined. For example to test that there is no selected Signal_anchor feature (somewhere in the sequence), use:

  case <AnyFeature:Signal_anchor<1> → selected feature must be absent

  Whereas without the operator, something like:

  case not <AnyFeature:Signal_anchor> → feature must be absent (whether selected or not)

• on a pattern being matched at a certain position of a GFF feature:

  case <Feature:PS50084:5=E> case <Feature:PS99999:10-13=N-{P}-[ST]-{P}>

• on the rule having been triggered by another rule:

  case <Triggered>

• on a metabolic pathway or property (only ‘=’ operator can be used):

  case <Property:Membrane=2> case <Property:NITROGEN_FIXATION> case <Property:NODULATION> case not <Property:METHANOG>

  Note (HAMAP ’hack’): If Property:Membrane is unknown (and if organism corresponding proteome is complete) it will be considered equal to 1 (so case <Property:Membrane=1> and case <Property:Membrane> will be true).

• on the presence of a (motif) feature with a certain InterPro id

  case <IPRO:IPR000859>

• on protein size

  case <Length<256> case <Length>420>

  Note: ProRule hack! to ’play’ with domain coverage...

5.1.1 Ternary logic

UniRule conditions should be evaluated using ternary logic, where conditions evaluate to one of three values: true, false, or undef. Operators are defined as follows, consistently with their implementation in the Perl programming language. Note that certain rules are counterintuitive.

Binary operators: ‘and’ and ‘or’

Unary operators: ‘not’ and ‘defined’

Operator associativity and precedence

The precedence order from highest to lowest and associativity are as follows.

Example of application: if the number of membranes is known and equal to 2, then apply a given annotation item. Otherwise, apply a less specific annotation item.

case defined <Property:Membrane> and <Property:Membrane=2>

CC   -!- SUBCELLULAR LOCATION: Inner membrane-associated (By similarity).

else

CC   -!- SUBCELLULAR LOCATION: Membrane-associated (By similarity).

end case

5.1.2 Condition line c! or c?

The condition line c! or c? contains additional constraints for the propagation of the line immediately below it. The format of this line is either:

c!   condition

c?   condition

where condition has the same syntax as in a case line, or, before a FT line, it can also include a PROSITE pattern expression.

Note: In FT lines, FT[Group] condition cannot be used (use case instead!).

The condition line differs from the case line in that

• the constraint affects only the next line (only 1 line!);
• it does not break a previous case condition.

The condition of the c! line must be true, otherwise an error is expected. Tools using UniRules are recommended to produce an error message.

Example:

c!   <Feature:PS00013>

KW   ATP-binding

The condition of the c? line may be true or not, as the feature does not appear in all matches of the UniRule.

Example:

c?   <Feature:PS99999:10-13=N-{P}-[ST]-{P}> and <OC:Eukaryota>

FT   CARBOHYD     10    13      N-linked (Potential).

Exceptions: See section Hidden information.

Transition: Condition lines should get automatically replaced by c! lines, and some of them later by c? lines. Mandatory conditions for disulfides should be suppressed, optional ones replaced by c? lines.

6. Hidden information

UniRules strive to include all information relevant to a motif. However, to avoid repetition, we did not include the information below, which is implicitly ‘known’ by the automatic annotation pipeline tools.

6.1 The keyword Repeat

The keyword Repeat is relevant to all rules of the data class Domain. This keyword applies when a domain or repeat is found at least twice in a protein. The corresponding part of the rule would be:

case <Feature:current_rule_accession_number>1>

KW   Repeat

end case

This behavior can be prevented by using the attribute ‘no keyword’ in the Repeats line (See section Repeats line).

6.2 The FT constraints line for the feature key DISULFID

For features with the key DISULFID, the constraint that the From and To positions both need to be a cysteine is implicit. The corresponding line would be the second line of the following example:

FT   DISULFID      4    23      By similarity.

FT   Condition: C-x*-C

A. UniAln

Introduction

UniAln is a format for protein sequence alignments that complement the UniRules collection. Some UniRules are developed based on predictors from specialized databases such as PROSITE. However, other UniRules are based on curated alignments that constitute the UniAln collection. That is the method used in the HAMAP annotation project.

Format

UniAln alignments are in a format similar to that produced by the CLUSTAL suite of programs. Each alignment is composed of:

• A Header line starting with the string ‘CLUSTAL’ or ‘MUSCLE’ or ‘T_COFFEE’. The rest of the line is free-text, but special strings are recognized by programs (see below).
• Two Blank lines.
• Alignment blocks, each one followed by a consensus line, separated by blank lines.

The alignment is subject to the following constraints:

• The file may not contain any trailing spaces at the end of lines. This allows safe editing with text editors that may suppress such spaces.
• The sequences in the alignment may only contain uppercase letters (with the exception of the letters ‘O’, ‘U’, ‘J’, ‘B’, ‘Z’), the gap character ‘-’, and the special characters ‘<’ and ‘>’ to indicate that a sequence portion has been excised to the NH2-side of the following amino acid, or to the COOH-side of the preceding amino acid, respectively. Note: although the letters ‘B’ and ‘Z’ are allowed by CLUSTAL, they are reserved internally to escape the special characters ‘<’ and ‘>’ when the alignment is converted through the CLUSTAL program.
• The line wrapping and content of the consensus lines must be such that the alignment may pass through the command ‘clustalw -convert’ without modification (after escaping the characters ‘<’ and ‘>’, and not taking into account trailing spaces at the end of consensus lines created by ‘clustalw -convert’).
• Sequence identifiers must be valid identifiers (ID line) from Swiss-Prot. The sequences must correspond to the sequences in Swiss-Prot. However, portions of sequences that are not aligned may be clipped by curators, and replaced with the special character ‘<’ (for N-terminal clipping) or ‘>’ (for C-terminal clipping), or by the succession ‘><’ (for internal clipping).

The alignment header line

The first line of the alignment must start with the string ‘CLUSTAL’ or ‘MUSCLE’ or ‘T_COFFEE’. The rest of the line is free-text, but special tags are recognized by programs. The tags may be repeated. Tags are:

‘template=identifier’

indicate that a sequence in the alignment is a template for feature propagation in the UniRule that uses the alignment. It is mandatory to indicate template sequences in alignments to allow alignment-based feature propagation in UniRules.

‘profile_method=method’

indicate the method that should be used to generate a profile from the alignment. Allowed values for method are:

‘profile_method=pfmake’

(default) A profile should be generated using ‘pfw’ and ‘pfmake’ from the PFTOOLS package. There is no need to indicate this method, as it is the default.

‘profile_method=hmmbuild’

A Hidden Markov Model should be generated using ‘hmmbuild’ from the HMMER package and converted to a profile using ‘htop’ from the PFTOOLS package. Profiles generated with ‘pfmake’ are usually more sensitive than those generated with ‘hmmbuild’. In some cases this means that they are less discriminating. If it is observed that the default method causes false positives, it can be attempted to use the ‘hmmbuild’ method to see if it solves the problem. See HAMAP 2003 paper for a discussion.

In a few HAMAP families, ‘hmmbuild’ was able to avoid false positives and negatives, while ‘pfmake’ was not:

• for closely related protein families;
• for certain very short proteins (‘pfmake’ gave very low scores).

Example header lines:

CLUSTAL

CLUSTAL W (1.83) multiple sequence alignment template=XYLA_ECOLI template=XYLA_ACTMI

CLUSTAL W (1.83) multiple sequence alignment template=XYLA_ECOLI profile_method=hmmbuild

MUSCLE (3.52) multiple sequence alignment

B. Sample UniRules entries

Example of a ’Domain’ UniRule

AC   PRU00241;

DC   Domain;

TR   PROSITE; PS50903; RUBREDOXIN_LIKE; 1; level=0

XX

Names: Rubredoxin-like domain

Function: It is involved in electron transfer processes.

XX

CC   -!- SIMILARITY: Contains # rubredoxin-like domain.

DR   PROSITE; PS00202; RUBREDOXIN; 0-1; trigger=no

case <FTGroup:1>

GO   GO:0009490; F:mononuclear iron electron carrier

GO   GO:0006810; P:transport

GO   GO:0006118; P:electron transport

KW   Transport

KW   Electron transport

KW   Metal-binding

KW   Iron

end case

XX

FT   From: PS50903

FT   DOMAIN     from     to       Rubredoxin-like #.

FT   METAL         6      6       Iron #1 (By similarity).

FT   Group: 1; Condition: C

FT   METAL         9      9       Iron #1 (By similarity).

FT   Group: 1; Condition: C

FT   METAL        38     38       Iron #1 (By similarity).

FT   Group: 1; Condition: C

FT   METAL        41     41       Iron #1 (By similarity).

FT   Group: 1; Condition: C

XX

Chop: Nter=0; Cter=0;

Size: 34-54;

Related: None;

Repeats: 2;

Topology: Cytoplasmic;

Example: Q9V099;

Scope:

 Bacteria

 Archaea

//

Example of a ’Protein’ UniRule

AC   MF_00198;

DC   Protein; auto

TR   HAMAP; MF_00198; -; 1; level=0

XX

ID   SPEE

case <OC:Bacteria>

DE   Spermidine synthase (EC 2.5.1.16) (Putrescine aminopropyltransferase)

DE   (PAPT) (SPDSY).

end case

case <OC:Archaea>

DE   Probable spermidine synthase (EC 2.5.1.16) (Putrescine

DE   aminopropyltransferase) (PAPT) (SPDSY).

end case

GN   Name=speE;

XX

CC   -!- FUNCTION: Catalyzes the production of spermidine from putrescine

CC       and decarboxylated S-adenosylmethionine (dcSAM), which serves as

CC       an aminopropyl donor (By similarity).

CC   -!- CATALYTIC ACTIVITY: S-adenosylmethioninamine + putrescine = 5'-S-

CC       methyl-5'-thioadenosine + spermidine.

CC   -!- PATHWAY: Amine and polyamine biosynthesis; spermidine

CC       biosynthesis; spermidine from putrescine: step 1/1.

case <OC:Proteobacteria>

CC   -!- SUBUNIT: Homodimer (By similarity).

else case <OC:Thermotogales>

CC   -!- SUBUNIT: Homotetramer (By similarity).

else

CC   -!- SUBUNIT: Homodimer or homotetramer (By similarity).

end case

CC   -!- SIMILARITY: Belongs to the spermidine/spermine synthase family.

XX

DR   Pfam; PF01564; Spermine_synth; 1; trigger=no

DR   TIGRFAMs; TIGR00417; speE; 1; trigger=no

DR   PROSITE; PS01330; SPERMIDINE_SYNTHASE_1; 1; trigger=no

DR   PROSITE; PS51006; SPERMIDINE_SYNTHASE_2; 1; trigger=no

XX

KW   Polyamine biosynthesis

KW   Spermidine biosynthesis

KW   Transferase

XX

GO   GO:0004766; F:spermidine synthase activity

GO   GO:0008295; P:spermidine biosynthetic process

XX

FT   From: SPEE_THEMA (Q9WZC2)

FT   REGION      152    153       S-adenosylmethioninamine binding (By

FT                                similarity).

FT   Condition: [DN]-[AGV]

FT   BINDING      46     46       S-adenosylmethioninamine (By similarity).

FT   Condition: [QHNR]

FT   BINDING     101    101       S-adenosylmethioninamine (By similarity).

FT   Condition: [DE]

FT   BINDING     121    121       S-adenosylmethioninamine (By similarity).

FT   Condition: [ED]

FT   BINDING     170    170       S-adenosylmethioninamine (By similarity).

FT   Condition: D

FT   BINDING     173    173       Putrescine (By similarity).

FT   Condition: [DE]

XX

Size: 261-366;

Related: None;

Template: P09158; P70998; Q9WZC2; Q8U4G1; O25503;

Scope:

 Bacteria

  not in AGRT5, ANASP, BACTN, BORBR, BORBU, BORPA, BORPE, BRAJA, BRUME,

  BRUSU, BUCBP, CAMJE, BLOFL, CAUCR, CHLCV, CHLMU, CHLPN, CHLTE, CHLTR,

  CORGL, COXBU, DEIRA, ENTFA, FUSNN, GLOVI, HAEDU, HAEIN, HELHP, LACLA,

  LACPL, LISIN, LISMO, MYCGA, MYCGE, MYCLE, MYCPE, MYCPN, MYCPU, PASMU,

  PORGI, PSEPK, RHILO, RHIME, RICCN, RICPR, STAAM, STAAN, STAAW, STAES,

  STRA3, STRA5, STRMU, STRP3, STRP8, STRP1, SYNEL, SYNY3, TREPA, TROW8,

  TROWT, UREPA, VIBCH, VIBPA, WIGBR

 Archaea

  not in HALSA, METAC, METKA, METMA, METTH

Fusion:

 NT: <Unknown>

 CT: <Unknown>

Duplicate: in AQUAE, BACAN, BACCR, LEPIN, PSEAE, RALSO, STRCO, THETN

Plasmid: in RALSO

Comments: None

**In Buchnera sp. only speE and speD are present, neither the pathway from ornithine

**nor the pathway from arginine are complete.

//

Table of Contents

About This Document

This document was generated using texi2html 1.82.

The buttons in the navigation panels have the following meaning:

where the Example assumes that the current position is at Subsubsection One-Two-Three of a document of the following structure:

• 1. Section One
  • 1.1 Subsection One-One
    • ...
  • 1.2 Subsection One-Two
    • 1.2.1 Subsubsection One-Two-One
    • 1.2.2 Subsubsection One-Two-Two
    • 1.2.3 Subsubsection One-Two-Three     <== Current Position
    • 1.2.4 Subsubsection One-Two-Four
  • 1.3 Subsection One-Three
    • ...
  • 1.4 Subsection One-Four