added fields for structural variants (#827)

* added fields for structural variants

* added sint32 type to protobuf to json-docs converter

* added other protobuf types to sphinx converter

Author: ejacox

python/ga4gh/schemas/_protocol_version.py

@@ -1,2 +1,2 @@
 # File generated by scripts/process_schemas.py; do not edit
-version = '0.6.0a9'
+version = '0.6.0a10'

python/ga4gh/schemas/ga4gh/variants_pb2.py

@@ -184,7 +184,7 @@ message Variant {
   // The bases that appear instead of the reference bases. Multiple alternate
   // alleles are possible.
   repeated string alternate_bases = 10;
-  
+
   // A map of additional information about the Variant.
   Attributes attributes = 13;
 
@@ -193,15 +193,40 @@ message Variant {
   // array are implicitly associated with this `Variant`.
   repeated Call calls = 12;
 
-  // True if filters were applied for this variant. VCF column 7 "FILTER" 
+  // The "variant_type" is used to denote e.g. structural variants.
+  // Examples:
+  //   DUP  : duplication of sequence following "start"; not necessarily in situ
+  //   DEL  : deletion of sequence following "start"
+  string variant_type = 17;
+
+  // Length of the - if labeled as such in variant_type - structural variation.
+  // Based on the use in VCFv4.2
+  int64 svlen = 18;
+
+  // In the case of structural variants, start and end of the variant may not
+  // be known with an exact base position. "cipos" provides an interval with
+  // high confidence for the start position. The interval is provided by 0 or
+  // 2 signed integers which are added to the start position.
+  // Based on the use in VCFv4.2
+  // Example:
+  //   [ -12000, 1000 ]
+  repeated sint32 cipos = 19;
+
+  // Similar to "cipos", but for the variant's end position (which is derived
+  // from start + svlen).
+  // Example:
+  //   [ -1000, 0 ]
+  repeated sint32 ciend = 20;
+
+  // True if filters were applied for this variant. VCF column 7 "FILTER"
   // any value other than the missing value.
   bool filters_applied = 14;
 
-  // True if all filters for this variant passed. VCF column 7 "FILTER" 
+  // True if all filters for this variant passed. VCF column 7 "FILTER"
   // value PASS.
   bool filters_passed = 15;
 
-  // Zero or more filters that failed for this variant. VCF column 7 "FILTER" 
+  // Zero or more filters that failed for this variant. VCF column 7 "FILTER"
   // shared across all alleles in the same VCF record.
   repeated string filters_failed = 16;
 }

src/main/proto/ga4gh/variants.proto

@@ -148,10 +148,15 @@ def type_to_string(f, map_types):
         return "uint64"
     elif f.type in [5]:
         return "integer"
+    elif f.type in [6]:
+        return "fixed64"
+    elif f.type in [7]:
+        return "fixed32"
     elif f.type in [8]:
         return "boolean"
     elif f.type in [9]:
         return "string"
+    # missing type 10 - Group
     elif f.type in [11, 14]:
         ref_name = f.ref_type
         if ref_name in map_types:
@@ -169,6 +174,16 @@ def type_to_string(f, map_types):
                 return kind
     elif f.type in [12]:
         return "bytes"
+    elif f.type in [13]:
+        return "uint32"
+    elif f.type in [15]:
+        return "sfixed32"
+    elif f.type in [16]:
+        return "sfixed64"
+    elif f.type in [17]:
+        return "sint32"
+    elif f.type in [18]:
+        return "sint64"
     else:
         raise Exception, f.type