Bio-LarK CR is a Human Phenotype Ontology (HPO) concept recognition tool. It uses a two-step approach to index and retrieve HPO concepts, combined with a series of language techniques to enable term normalization. In addition to standard CR, Bio-LarK CR is able to decompose and align conjunctive terms, of the form 'short and broad fingers', as well as recognise and process non-canonical phenotypes - e.g., 'fingers are short and broad'.
Bio-LarK CR is available via a REST API at: http://bio-lark.org:8080/biolark/annotate
Input parameters:
text
- the text to be annotated.dataSource
- currently only Human Phenotype Ontology
is supported.http://bio-lark.org:8080/biolark/annotate?text=The long bones are shortened, the metaphyses flared,
cupped and irregular but the density and structure of the bones is grossly normal.&dataSource=Human Phenotype Ontology
Response: A JSON-formatted list of concepts:
dataSource
- currently only Human Phenotype Ontology
startOffset
- absolute start index of the annotated text span.endOffset
- absolute end index of the annotated text span.originalSpan
- the actual text span being annotated.uri
- the HPO concept assigned to the annotated text span.The call listed above will result in the following reponse:
[
{
"dataSource": "Human Phenotype Ontology",
"startOffset": "34",
"endOffset": "51",
"originalSpan": "metaphyses flared",
"uri": "http://purl.obolibrary.org/obo/HP_0003015"
},
{
"dataSource": "Human Phenotype Ontology",
"startOffset": "4",
"endOffset": "28",
"originalSpan": "long bones shortened",
"uri": "http://purl.obolibrary.org/obo/HP_0003026"
},
{
"dataSource": "Human Phenotype Ontology",
"startOffset": "143",
"endOffset": "152",
"originalSpan": "back pain",
"uri": "http://purl.obolibrary.org/obo/HP_0003418"
},
{
"dataSource": "Human Phenotype Ontology",
"startOffset": "148",
"endOffset": "152",
"originalSpan": "pain",
"uri": "http://purl.obolibrary.org/obo/HP_0012531"
}
]