cgcFileSizeByDisease.rq

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
PREFIX tcga: <https://www.sbgenomics.com/ontologies/2014/11/tcga#> 
 
SELECT ?inv_label (count(?file_size) as ?N_files) (sum(?file_size)*0.000000000931323 as ?N_GiB) WHERE 
{ 
  
  ?file a tcga:File .
  ?file rdfs:label ?file_name .
  
  ?file tcga:hasSample ?sample .
  ?sample tcga:hasSampleType ?st .
  ?st rdfs:label ?sample_type_label 
  filter(?sample_type_label='Primary Tumor' || ?sample_type_label='Additional - New Primary' || ?sample_type_label='Additional Metastatic' || ?sample_type_label='Metastatic' || ?sample_type_label='Primary Blood Derived Cancer - Peripheral Blood' || ?sample_type_label='Recurrent Tumor' || ?sample_type_label='Blood Derived Normal' || ?sample_type_label='Bone Marrow Normal' || ?sample_type_label='Buccal Cell Normal' || ?sample_type_label='Solid Tissue Normal')
  
  
  ?file tcga:hasExperimentalStrategy ?xs .
  ?xs rdfs:label ?experimental_strategy_label .
  filter(?experimental_strategy_label='RNA-Seq' || ?experimental_strategy_label='WGS') .
  
  ?file tcga:hasDataType ?type .
  ?type rdfs:label ?data_type_label .
  filter(?data_type_label='Raw sequencing data') .
  
  ?file tcga:hasDataFormat ?format .
  ?format rdfs:label ?data_format_label .
  filter(?data_format_label='BAM') .
  
  ?file a tcga:File .
  ?file tcga:hasSize ?file_size .

  ?file tcga:hasCase ?case .
  ?case tcga:hasInvestigation ?investigation .
  ?investigation rdfs:label ?inv_label .
} 

GROUP BY ?inv_label
ORDER BY ?N_GiB