Varquery can technically query any JSON data which it has indexed. Though it is designed with VCF atomized JSON in mind. / is used as a kind of depth-based delimiter. i.e
{"key1":"A", "key2":{"key3":"B"}}
/key1 is querying the JSON root object for key1
/key2/key3 is querying the JSON "key2" object for key3
Generally you can query data like so:
// basic equals queries with operators
/key = val
/key1 = val1 AND /key2 = val2
/key1 = val1 OR /key2 = val2
(/key1 = val1 OR /key2 = val2) AND /key3 = val3
NOT ((/key1 = val1 OR /key2 = val2) AND /key3 = val3)
// query multipe values with operators
/key1 = ( val1 OR val2) AND /key3 = val3
// Integer Numeric queries
/key1 == 1
/key1 = 1:3
/key1 > 1
/key1 >= 1
/key1 < 1
/key1 <= 1
// float Numeric queries
/key1 == 1f
/key1 = 1f:3f
/key1 > 1f
/key1 >= 1f
/key1 < 1f
/key1 <= 1f
Of note for VCF fields, since they are record oriented but have sub fields we represent this as such:
/INFO/AF > 0.5f
/FORMAT/AF > 0.5f
/FORMAT/GT = ./.
The atomize_vcf program supports the parsing of annotation fields which in VCF are often pipe-delimited (|). This allows the ability to search/filter records based on annotation information. Currently we only support SnpEff's ANN field. This will change.
Since an ANN field can have multiple annotations this data is atomized into an array of objects in JSON.
This VCF field (not a real annotation, it has been modified for this example) | Second annotation starts here
ANN = A|intron_variant|MODIFIER|PLCXD1|ENSG00000182378|Transcript|ENST00000381657|protein_coding||1/6|ENST00000381657.2:c.-21-26C>A|||||,A|missense_variant|MODIFIER|PLCXD1|ENSG00000182378|Transcript|ENST00000381663|protein_coding||1/7|ENST00000381663.3:c.-21-26C>A|||||
becomes this JSON, blank fields are not reported
"ANN" : [{"allele" : "A", "effect" : "intron_variant", ...},{"allele" : "A", "effect" : "missense_variant", ...}]
// SNPEFF ANN field names
"allele",
"effect",
"impact",
"gene_name",
"gene_id",
"feature_type",
"feature_id",
"transcript_biotype",
"rank_total",
"hgvs_c",
"hgvs_p",
"cdna_position",
"cds_position",
"protein_position",
"distance_to_feature",
"errors_warnings_info"
Keys and values cannot have any whitespace. This functionality will be added.