POST query support multiple value combinations from multiple fields

[newgene]

here is a use case from MyVariant.info: biothings/myvariant.info#89.

Possible solution:

q="1|12345,2|1000", scopes="chrom|vcf.position"

Using "|" to separate the value combination in q and the corresponding field combination in scopes.

A few other notes:

• "|" may not be the best choice, need to pick one won't be used in the value itself and the query separator.
• The number of value/field combination must match.
• Should set a cap for maximium number of value/field combinations (10 should be good enough)
• should be 100% back compatible with the current behavior
• can still have multiple fields to match one value, i.e., this should work: q=val1|val2&scopes=fld1,fld2|fld3, where val1 should match either fld1 or fld2.

[namespacestd0]

Syntax shown in an example here:

{
  "q": [["CDK2", "c.314A>T"], ["CXCR4", "c.535G>C"]],
  "scopes": [["dbsnp.gene.symbol", "snpeff.ann.genename"], "snpeff.ann.hgvs_c"]
}

will basically be equivalent to

[
    {
        "query": {
            "bool": {
                "must": [
                    {
                        "multi_match": {
                            "query": "CDK2",
                            "operator": "and",
                            "fields": [
                                "dbsnp.gene.symbol",
                                "snpeff.ann.genename"
                            ]
                        }
                    },
                    {
                        "multi_match": {
                            "query": "c.314A>T",
                            "operator": "and",
                            "fields": "snpeff.ann.hgvs_c"
                        }
                    }
                ]
            }
        }
    },
    {
        "query": {
            "bool": {
                "must": [
                    {
                        "multi_match": {
                            "query": "CXCR4",
                            "operator": "and",
                            "fields": [
                                "dbsnp.gene.symbol",
                                "snpeff.ann.genename"
                            ]
                        }
                    },
                    {
                        "multi_match": {
                            "query": "c.535G>C",
                            "operator": "and",
                            "fields": "snpeff.ann.hgvs_c"
                        }
                    }
                ]
            }
        }
    }
]

Implemented in 89c0e0c
Tests/examples in 996420a
Myvariant use cases/tests in biothings/myvariant.info@fc7fac7
Pending deployment.

[kevinxin90]

@namespacestd0 Is this feature deployed? Tried on myvariant, seems not working at the moment.

[namespacestd0]

This feature should be universally available on all major biothings instances now.

[newgene]

for reference, these two tests provide some example queries:

biothings.api/tests/web/handlers/test_query.py

Lines 780 to 828 in 6dce4ce

	def test_20_nested(self):
	"""
	[
	{
	"query": [
	"cdk2",
	"9555"
	],
	"_id": "101025892",
	"_score": 1.0840356,
	"entrezgene": "101025892",
	"name": "cyclin dependent kinase 2",
	"symbol": "CDK2",
	"taxid": 9555
	}
	]
	"""
	payload = {
	"q": [["cdk2", "9555"]],
	"scopes": ["symbol", "taxid"],
	"fields": ["symbol", "name", "taxid", "entrezgene", "ensemblgene"]
	}
	ans = self.query(method='POST', json=payload)
	assert len(ans) == 1
	def test_21_nested(self):
	"""
	[
	{
	"query": [
	"101025892",
	"9555"
	],
	"_id": "101025892",
	"_score": 3.8134108,
	"entrezgene": "101025892",
	"name": "cyclin dependent kinase 2",
	"symbol": "CDK2",
	"taxid": 9555
	}
	]
	"""
	payload = {
	"q": [["101025892", "9555"]],
	"scopes": [["symbol", "entrezgene"], "taxid"],
	"fields": ["symbol", "name", "taxid", "entrezgene", "ensemblgene"]
	}
	ans = self.query(method='POST', json=payload)
	assert len(ans) == 1