# Counts of tissue-specific genes and proteins per tissue
# Count genes and proteins identified as specific to each tissue (using TiSMeD's own tissue classes).
# Excludes counts from subtissues.

PREFIX tismedo: <http://med2rdf.org/tismed/ontology#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT ?tissue ?tissue_label
       (COALESCE(?g, 0) AS ?gene_count)
       (COALESCE(?p, 0) AS ?protein_count)
WHERE {
  ?tissue rdfs:label ?tissue_label .
  OPTIONAL {
    SELECT ?tissue (COUNT(DISTINCT ?gene) AS ?g) WHERE {
      ?gene tismedo:hasGeneExpressionSpecificity/tismedo:isPositivelySpecificTo ?tissue .
    } GROUP BY ?tissue
  }
  OPTIONAL {
    SELECT ?tissue (COUNT(DISTINCT ?protein) AS ?p) WHERE {
      ?protein tismedo:hasProteinExpressionSpecificity/tismedo:isPositivelySpecificTo ?tissue .
    } GROUP BY ?tissue
  }
  FILTER (BOUND(?g) || BOUND(?p))
}
ORDER BY DESC(?gene_count)

# Retrieve tissues with specific expression of a given gene and their scores using the gene symbol
# Example: INS (insulin). Gene symbols are stored in rdfs:label.

PREFIX tismedo: <http://med2rdf.org/tismed/ontology#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT ?gene ?tissue ?tissue_label ?uberon
       ?mean_spm ?t_score ?dataset_count ?ts_protein
WHERE {
  VALUES ?symbol { "INS" }
  ?gene rdfs:label ?symbol ;
        tismedo:hasGeneExpressionSpecificity ?spec .
  ?spec tismedo:isPositivelySpecificTo ?tissue ;
        tismedo:meanSpmAdjustValue ?mean_spm ;
        tismedo:tScore ?t_score ;
        tismedo:supportDatasetCount ?dataset_count ;
        tismedo:hasTissueSpecificProtein ?ts_protein .
  ?tissue rdfs:label ?tissue_label .
  OPTIONAL { ?tissue skos:closeMatch ?uberon }
}
ORDER BY DESC(?t_score)

# Retrieve genes specific to a given tissue or any of its subtissues, ordered by confidence from highest to lowest
# The tissue hierarchy is traversed using rdfs:subClassOf*, so specifying "brain", for example,
# also includes genes specific to subtissues such as "brain_amygdala".

PREFIX tismedo: <http://med2rdf.org/tismed/ontology#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT ?gene ?symbol ?tissue_label ?mean_spm ?t_score ?dataset_count
WHERE {
  VALUES ?top_label { "brain" }
  ?top rdfs:label ?top_label .
  ?tissue rdfs:subClassOf* ?top ;
          rdfs:label ?tissue_label .
  ?spec tismedo:isPositivelySpecificTo ?tissue ;
        tismedo:meanSpmAdjustValue ?mean_spm ;
        tismedo:tScore ?t_score ;
        tismedo:supportDatasetCount ?dataset_count .
  ?gene tismedo:hasGeneExpressionSpecificity ?spec ;
        rdfs:label ?symbol .
}
ORDER BY DESC(?t_score) DESC(?dataset_count) DESC(?mean_spm)
LIMIT 100

# High-confidence genes with tissue specificity also confirmed at the protein level
# The thresholds are set more stringently than TiSMeD's criteria
# (meanSPM >= 0.8, T-score >= 0.6, number of datasets >= 2). Adjust them as needed for your use case.

PREFIX tismedo: <http://med2rdf.org/tismed/ontology#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT ?gene ?symbol ?tissue_label ?mean_spm ?t_score ?dataset_count
WHERE {
  ?gene tismedo:hasGeneExpressionSpecificity ?spec ;
        rdfs:label ?symbol .
  ?spec tismedo:hasTissueSpecificProtein true ;
        tismedo:isPositivelySpecificTo ?tissue ;
        tismedo:meanSpmAdjustValue ?mean_spm ;
        tismedo:tScore ?t_score ;
        tismedo:supportDatasetCount ?dataset_count .
  ?tissue rdfs:label ?tissue_label .
  FILTER (?mean_spm >= 0.95 && ?t_score >= 0.9 && ?dataset_count >= 5)
}
ORDER BY ?tissue_label DESC(?dataset_count)

# Find TiSMeD tissues by UBERON ID and retrieve genes specific to those tissues
# This can serve as a starting point for integration with other UBERON-based datasets (e.g., Bgee and GTEx).
# Example: UBERON:0002107 (liver)

PREFIX tismedo: <http://med2rdf.org/tismed/ontology#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
PREFIX obo: <http://purl.obolibrary.org/obo/>

SELECT ?uberon ?tissue ?tissue_label ?gene ?symbol ?t_score
WHERE {
  VALUES ?uberon { obo:UBERON_0002107 }
  ?tissue skos:closeMatch ?uberon ;
          rdfs:label ?tissue_label .
  ?gene tismedo:hasGeneExpressionSpecificity ?spec ;
        rdfs:label ?symbol .
  ?spec tismedo:isPositivelySpecificTo ?tissue ;
        tismedo:tScore ?t_score .
}
ORDER BY DESC(?t_score)