SIB SPARQL training

📓 Download this page as a Jupyter notebook

Sequences & isoforms

This page shows you how UniProtKB stores protein sequences and their isoforms. These queries are adapted from the SIB SPARQL examples collection for UniProt.

A UniProt entry points at its sequence(s) through the up:sequence property, and, when a computational mapping exists, through up:potentialSequence as well. Each sequence is its own resource - typically an isoform, identified by an IRI like http://purl.uniprot.org/isoforms/P05067-1 - carrying the residues themselves in rdf:value. A sequence resource is typed up:Simple_Sequence when UniProt maintains it directly, or up:External_Sequence when it comes from an external isoform mapping.

Retrieving sequences for an organism

Adapted from sparql-examples UniProt/3.

Select UniProtKB entries and their amino acid sequences (including isoforms) for E. coli K12 and all its strains.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix taxon: <http://purl.uniprot.org/taxonomy/>
prefix isoform: <http://purl.uniprot.org/isoforms/>

taxon:511145 rdfs:label "Escherichia coli str. K-12 substr. MG1655" ;
             rdfs:subClassOf taxon:83333 .

<P0A877> a up:Protein ;
  up:organism taxon:511145 ;
  up:sequence isoform:P0A877-1 .

isoform:P0A877-1 rdf:value "MTEQAPAHWDIKDFAKVDQKALTHSVRKVASQVLGISPD" .
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX taxon: <http://purl.uniprot.org/taxonomy/>
PREFIX up: <http://purl.uniprot.org/core/>

SELECT ?protein ?organism ?isoform ?sequence
WHERE
{
    ?protein a up:Protein .
    ?protein up:organism ?organism .
    # Taxon subclasses are materialized, do not use rdfs:subClassOf+
    ?organism rdfs:subClassOf taxon:83333 .
    ?protein up:sequence ?isoform .
    ?isoform rdf:value ?sequence .
}

Computationally mapped isoforms

Adapted from sparql-examples UniProt/101.

Some isoforms aren’t asserted directly by UniProt curators, but computationally mapped instead, through up:potentialSequence. On the real endpoint, sequence queries are often scoped to the http://sparql.uniprot.org/uniprot named graph, so they don’t also match data living in the separate UniParc graph. A single in-page example dataset has only one graph to begin with, so that wrapper can simply be dropped.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix taxon: <http://purl.uniprot.org/taxonomy/>
prefix isoform: <http://purl.uniprot.org/isoforms/>

<P05067> a up:Protein ;
  up:organism taxon:9606 ;
  up:potentialSequence isoform:P05067-11 .

isoform:P05067-11 a up:External_Sequence .
PREFIX taxon: <http://purl.uniprot.org/taxonomy/>
PREFIX up: <http://purl.uniprot.org/core/>

SELECT ?entry ?sequence
WHERE {
  ?entry a up:Protein ;
    up:organism taxon:9606 ;
    up:potentialSequence ?sequence .
}

Which sequence is canonical?

Adapted from sparql-examples UniProt/107.

A UniProt entry can have several isoforms, but up:sequence always points at whichever one is canonical. This query marks each sequence as canonical or not: a Simple_Sequence is likely canonical, unless it’s also an External_Sequence whose IRI doesn’t correspond to the entry’s own accession (an external isoform mapped in from elsewhere).

Comunica limitation

Both queries below use a FILTER inside an OPTIONAL block that depends on a value from a BIND - a combination Comunica (the engine running the examples on this site) can’t plan when evaluating a query itself against a small local dataset, even though it’s valid SPARQL. Sent whole to a real, conformant endpoint instead - which is exactly what happens below, since the query targets sparql.uniprot.org live, with no local fixture - that endpoint does its own native evaluation and there’s no issue.

This query runs live against the real public endpoint https://sparql.uniprot.org/sparql — there's no local example data here, so it makes a real network request and results (and response time) depend on that service.

PREFIX taxon: <http://purl.uniprot.org/taxonomy/>
PREFIX up: <http://purl.uniprot.org/core/>

SELECT ?entry ?sequence ?isCanonical
WHERE {
  GRAPH <http://sparql.uniprot.org/uniprot> {
      ?entry a up:Protein ;
	up:organism taxon:9606 ;
      	up:sequence ?sequence .
      # If the sequence is a "Simple_Sequence" it is likely to be the
      # cannonical sequence
      OPTIONAL {
       	?sequence a up:Simple_Sequence .
        BIND(true AS ?likelyIsCanonical)
      }
      # unless we are dealing with an external isoform
      # see https://www.uniprot.org/help/canonical_and_isoforms
      OPTIONAL {
       	FILTER(?likelyIsCanonical)
        ?sequence a up:External_Sequence .
        BIND(true AS ?isComplicated)
      }
      # If it is an external isoform it's id would not match the
      # entry primary accession. We test that the variable is bound
      # else an UNDEF might fail the query from working
      BIND(IF(BOUND(?isComplicated), STRENDS(STR(?entry), STRBEFORE(SUBSTR(STR(?sequence), 34),'-')),?likelyIsCanonical) AS ?isCanonical)
  }
}
LIMIT 5

The canonical isoform doesn’t have to end in “-1”

Adapted from sparql-examples UniProt/163.

up:sequence points at whichever isoform was chosen as canonical - usually -1, but not always. This query finds reviewed proteins where it isn’t, ordered by the highest isoform number used as the canonical sequence. Like the query above, this one also mixes a FILTER inside an OPTIONAL with a BIND, so it runs live against sparql.uniprot.org rather than a local fixture.

This query runs live against the real public endpoint https://sparql.uniprot.org/sparql — there's no local example data here, so it makes a real network request and results (and response time) depend on that service.

PREFIX up: <http://purl.uniprot.org/core/>
PREFIX uniprotkb: <http://purl.uniprot.org/uniprot/>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>

SELECT
  ?protein
  (?sequence AS ?canonicalIsoform)
  ?isoformCount
WHERE
{
    ?protein up:reviewed true .
    ?protein up:sequence ?sequence .
    ?sequence a up:Simple_Sequence .
    OPTIONAL {
      ?sequence a up:External_Sequence .
      BIND(true AS ?isExternalSequence)
      BIND(SUBSTR(STR(?protein), STRLEN(STR(uniprotkb:))) AS ?proteinAc)
      FILTER(CONTAINS(STR(?sequence), ?proteinAc))
    }
    FILTER(?isExternalSequence || !BOUND(?isExternalSequence))
  BIND(xsd:int(STRAFTER(STR(?sequence), "-")) AS ?isoformCount)
} ORDER BY DESC(?isoformCount)
LIMIT 5

Initiator methionine

Translation always starts with a methionine, but that residue is often enzymatically removed afterwards. When UniProt has evidence for this, it’s recorded as a up:Initiator_Methionine_Annotation - a single-residue FALDO position (faldo:begin and faldo:end both point at the same spot) at the very start of the sequence. Below, human hemoglobin subunit alpha (P69905): the query looks up that position in the full (unprocessed) sequence to double-check it really is an “M”.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix faldo: <http://biohackathon.org/resource/faldo#>
prefix isoform: <http://purl.uniprot.org/isoforms/>

<P69905> a up:Protein ;
  up:sequence isoform:P69905-1 ;
  up:annotation <P69905#InitMet> .

isoform:P69905-1 rdf:value "MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHFDLSHGSAQVKGHGKKVADALTNAVAHVDDMPNALSALSDLHAHKLRVDPVNFKLLSHCLLVTLAAHLPAEFTPAVHASLDKFLASVSTVLTSKYR" .

<P69905#InitMet> a up:Initiator_Methionine_Annotation ;
  up:range <P69905#InitMet_range> .

<P69905#InitMet_range> faldo:begin <P69905#pos1> ;
  faldo:end <P69905#pos1> .

<P69905#pos1> faldo:position 1 ;
  faldo:reference isoform:P69905-1 .
PREFIX faldo: <http://biohackathon.org/resource/faldo#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX up: <http://purl.uniprot.org/core/>

SELECT
  ?protein
  ?removedResidue
WHERE {
  ?protein up:annotation ?met ;
    up:sequence ?sequence .
  ?met a up:Initiator_Methionine_Annotation ;
    up:range/faldo:begin ?begin .
  ?begin faldo:position ?position ;
    faldo:reference ?sequence .
  ?sequence rdf:value ?sequenceVal .
  BIND(SUBSTR(?sequenceVal, ?position, 1) AS ?removedResidue)
}

Chains: the mature, processed protein

Once initiator methionines, signal peptides and other processing steps are accounted for, what’s left is the mature protein - recorded as a up:Chain_Annotation with a rdfs:comment naming it and a range covering the residues it spans. Continuing the hemoglobin example: after the initiator methionine above is removed, the mature chain covers residues 2–142 of the same 142-residue sequence.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix faldo: <http://biohackathon.org/resource/faldo#>
prefix isoform: <http://purl.uniprot.org/isoforms/>

<P69905> a up:Protein ;
  up:sequence isoform:P69905-1 ;
  up:annotation <P69905#Chain1> .

isoform:P69905-1 rdf:value "MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTTKTYFPHFDLSHGSAQVKGHGKKVADALTNAVAHVDDMPNALSALSDLHAHKLRVDPVNFKLLSHCLLVTLAAHLPAEFTPAVHASLDKFLASVSTVLTSKYR" .

<P69905#Chain1> a up:Chain_Annotation ;
  rdfs:comment "Hemoglobin subunit alpha" ;
  up:range <P69905#Chain1_range> .

<P69905#Chain1_range> faldo:begin <P69905#pos2> ;
  faldo:end <P69905#pos142> .

<P69905#pos2> faldo:position 2 ;
  faldo:reference isoform:P69905-1 .

<P69905#pos142> faldo:position 142 ;
  faldo:reference isoform:P69905-1 .
PREFIX faldo: <http://biohackathon.org/resource/faldo#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX up: <http://purl.uniprot.org/core/>

SELECT
  ?protein
  ?description
  ?begin
  ?end
WHERE {
  ?protein up:annotation ?chain .
  ?chain a up:Chain_Annotation ;
    rdfs:comment ?description ;
    up:range ?range .
  ?range faldo:begin/faldo:position ?begin ;
    faldo:end/faldo:position ?end .
}

Signal peptides

Secreted and membrane proteins carry a up:Signal_Peptide_Annotation at their N-terminus - a short stretch that targets the protein for translocation and is cleaved off before the mature chain begins, so it’s never part of the functional protein. Below, the amyloid precursor protein (P05067) again, this time extracting the actual signal peptide sequence (residues 1–17) with SUBSTR.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix faldo: <http://biohackathon.org/resource/faldo#>
prefix isoform: <http://purl.uniprot.org/isoforms/>

<P05067> a up:Protein ;
  up:sequence isoform:P05067-1 ;
  up:annotation <P05067#Signal1> .

isoform:P05067-1 rdf:value "MLPGLALLLLAAWTARALEVPTDGNAGLL" .

<P05067#Signal1> a up:Signal_Peptide_Annotation ;
  up:range <P05067#Signal1_range> .

<P05067#Signal1_range> faldo:begin <P05067#pos1> ;
  faldo:end <P05067#pos17> .

<P05067#pos1> faldo:position 1 ;
  faldo:reference isoform:P05067-1 .

<P05067#pos17> faldo:position 17 ;
  faldo:reference isoform:P05067-1 .
PREFIX faldo: <http://biohackathon.org/resource/faldo#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX up: <http://purl.uniprot.org/core/>

SELECT
  ?protein
  ?signalPeptide
WHERE {
  ?protein up:annotation ?sp ;
    up:sequence ?sequence .
  ?sp a up:Signal_Peptide_Annotation ;
    up:range ?range .
  ?range faldo:begin/faldo:position ?begin ;
    faldo:end/faldo:position ?end .
  ?sequence rdf:value ?sequenceVal .
  BIND(SUBSTR(?sequenceVal, ?begin, ?end - ?begin + 1) AS ?signalPeptide)
}

Fragmented sequences

Adapted from sparql-examples UniProt/41.

Not every sequence is complete: up:fragment marks a sequence as being composed of fragments.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix isoform: <http://purl.uniprot.org/isoforms/>

<P0A877> a up:Protein ;
  up:sequence isoform:P0A877-1 .
isoform:P0A877-1 up:fragment [ a up:Fragment ] .

<P05067> a up:Protein ;
  up:sequence isoform:P05067-1 .
isoform:P05067-1 rdf:value "MTEQAP" .
PREFIX up: <http://purl.uniprot.org/core/>

SELECT DISTINCT
  ?protein
WHERE {
  ?protein a up:Protein ;
    up:sequence ?sequence .
  ?sequence up:fragment [] .
}

Mass spectrometry measurements

Adapted from sparql-examples UniProt/210.

Some entries carry an experimentally measured mass, recorded as a up:Mass_Spectrometry_Annotation with a up:measuredValue and, where known, a up:measuredError margin.

Example data (Turtle) — edit it, then re-run any query below
base <http://purl.uniprot.org/uniprot/>
prefix up: <http://purl.uniprot.org/core/>
prefix xsd: <http://www.w3.org/2001/XMLSchema#>

<P05067> a up:Protein ;
  up:annotation <P05067#MS1> .
<P05067#MS1> a up:Mass_Spectrometry_Annotation ;
  up:measuredValue "79924.5"^^xsd:double ;
  up:measuredError "50.0"^^xsd:double .

<P0A877> a up:Protein ;
  up:annotation <P0A877#MS1> .
<P0A877#MS1> a up:Mass_Spectrometry_Annotation ;
  up:measuredValue "55000.0"^^xsd:double .
PREFIX up: <http://purl.uniprot.org/core/>

SELECT
  ?protein
  ?annotation
  ?measuredValue
  ?measuredError
WHERE {
  ?protein a up:Protein ;
    up:annotation ?annotation .
  ?annotation a up:Mass_Spectrometry_Annotation ;
    up:measuredValue ?measuredValue .
  OPTIONAL { ?annotation up:measuredError ?measuredError }
}