Merge pull request #613 from legumeinfo/genefamily_pangene_linkouts

added linkout services for gene families and pan gene sets.

linkouts/compose.prod.yml

@@ -1,6 +1,6 @@
 services:
   linkouts:
-    image: ghcr.io/legumeinfo/microservices-linkouts:1.0.0
+    image: ghcr.io/legumeinfo/microservices-linkouts:1.0.1
     ports:
       - "${PORT:-8080}:8080"
     volumes:

linkouts/linkouts/http_server.py

@@ -4,8 +4,12 @@
 
 GENES_PATH = "/gene_linkouts"
 GENOMIC_REGIONS_PATH = "/genomic_region_linkouts"
+GENE_FAMILIES_PATH = "/gene_family_linkouts"
+PAN_GENE_SETS_PATH = "/pan_gene_set_linkouts"
 GENES_QUERY = "genes"
 GENOMIC_REGIONS_QUERY = "genomic_regions"
+GENE_FAMILIES_QUERY = "gene_families"
+PAN_GENE_SETS_QUERY = "pan_gene_sets"
 
 
 async def http_genes_get_handler(request):
@@ -60,6 +64,58 @@ async def http_genomic_regions_post_handler(request):
     return web.json_response(linkouts)
 
 
+async def http_gene_families_get_handler(request):
+    # parse the query from the request query string
+    try:
+        ids = request.rel_url.query[GENE_FAMILIES_QUERY]
+    except KeyError:
+        raise web.HTTPBadRequest(text="No " + GENE_FAMILIES_QUERY + " supplied")
+    ids = ids.split(",")
+    handler = request.app["handler"]
+    linkouts = handler.process_gene_families(ids)
+    return web.json_response(linkouts)
+
+
+async def http_gene_families_post_handler(request):
+    # parse the query from the request POST data
+    data = await request.json()
+    ids = data.get(GENE_FAMILIES_QUERY, [])
+    if type(ids) != list:
+        raise web.HTTPBadRequest(text=GENE_FAMILIES_QUERY + " must be given as list")
+    if len(ids) == 0:
+        raise web.HTTPBadRequest(text="No " + GENE_FAMILIES_QUERY + " supplied")
+
+    handler = request.app["handler"]
+    linkouts = handler.process_gene_families(ids)
+    return web.json_response(linkouts)
+
+
+async def http_pan_gene_sets_get_handler(request):
+    # parse the query from the request query string
+    try:
+        ids = request.rel_url.query[PAN_GENE_SETS_QUERY]
+    except KeyError:
+        raise web.HTTPBadRequest(text="No " + PAN_GENE_SETS_QUERY + " supplied")
+    ids = ids.split(",")
+    handler = request.app["handler"]
+    linkouts = handler.process_pan_gene_sets(ids)
+    return web.json_response(linkouts)
+
+
+async def http_pan_gene_sets_post_handler(request):
+    # parse the query from the request POST data
+    data = await request.json()
+    ids = data.get(PAN_GENE_SETS_QUERY, [])
+    if type(ids) != list:
+        raise web.HTTPBadRequest(text=PAN_GENE_SETS_QUERY + " must be given as list")
+    if len(ids) == 0:
+        raise web.HTTPBadRequest(text="No " + PAN_GENE_SETS_QUERY + " supplied")
+
+    handler = request.app["handler"]
+    linkouts = handler.process_pan_gene_sets(ids)
+    return web.json_response(linkouts)
+
+
 def run_http_server(host, port, handler):
     # make the app
     app = web.Application()
@@ -83,6 +139,14 @@ def run_http_server(host, port, handler):
     cors.add(route)
     route = app.router.add_get(GENOMIC_REGIONS_PATH, http_genomic_regions_get_handler)
     cors.add(route)
+    route = app.router.add_post(GENE_FAMILIES_PATH, http_gene_families_post_handler)
+    cors.add(route)
+    route = app.router.add_get(GENE_FAMILIES_PATH, http_gene_families_get_handler)
+    cors.add(route)
+    route = app.router.add_post(PAN_GENE_SETS_PATH, http_pan_gene_sets_post_handler)
+    cors.add(route)
+    route = app.router.add_get(PAN_GENE_SETS_PATH, http_pan_gene_sets_get_handler)
+    cors.add(route)
     # run the app
     web.run_app(app)
     # TODO: what about teardown? runner.cleanup()

linkouts/linkouts/request_handler.py

@@ -10,6 +10,7 @@
 
 GENE_LINKOUTS = "gene_linkouts"
 GENE_REGEX = re.compile("{GENE_ID}")
+UNPREFIXED_GENE_REGEX = re.compile("{UNPREFIXED_GENE_ID}")
 
 GENOMIC_REGION_LINKOUTS = "genomic_region_linkouts"
 GENOMIC_REGION_SPECIFIER_REGEX = re.compile(r"^([^:]+):(\d+)(-|\.\.)(\d+)$")
@@ -18,6 +19,12 @@
 GENOMIC_REGION_START_REGEX = re.compile("{GENOMIC_REGION_START}")
 GENOMIC_REGION_END_REGEX = re.compile("{GENOMIC_REGION_END}")
 
+GENE_FAMILY_LINKOUTS = "gene_family_linkouts"
+GENE_FAMILY_REGEX = re.compile("{GENE_FAMILY_ID}")
+
+PAN_GENE_SET_LINKOUTS = "pan_gene_set_linkouts"
+PAN_GENE_SET_REGEX = re.compile("{PAN_GENE_SET_ID}")
+
 
 class RequestHandler:
     def __init__(self, lglob_root):
@@ -36,7 +43,12 @@ def _read_linkouts(self, lfile):
         yml = yaml.load(f.read(), Loader=yaml.FullLoader)
         prefix = yml["prefix"]
 
-        for linkable_type in [GENE_LINKOUTS, GENOMIC_REGION_LINKOUTS]:
+        for linkable_type in [
+            GENE_LINKOUTS,
+            GENOMIC_REGION_LINKOUTS,
+            GENE_FAMILY_LINKOUTS,
+            PAN_GENE_SET_LINKOUTS,
+        ]:
             if yml.get(linkable_type) is not None:
                 for linkout in yml[linkable_type]:
                     if self.linkout_lookup.get(linkable_type) is None:
@@ -55,6 +67,7 @@ def process_genes(self, ids):
 
         for id in ids:
             prefix = ".".join(id.split(".")[0:4])
+            unprefixed_id = ".".join(id.split(".")[4:])
             if type_lookup.get(prefix) is not None:
                 templates = type_lookup[prefix]
                 for template in templates:
@@ -63,7 +76,13 @@ def process_genes(self, ids):
                     # TODO: if method is POST, we probably need to do something with the
                     # request body content
                     linkout[HREF] = GENE_REGEX.sub(id, template[HREF])
+                    linkout[HREF] = UNPREFIXED_GENE_REGEX.sub(
+                        unprefixed_id, linkout[HREF]
+                    )
                     linkout[TEXT] = GENE_REGEX.sub(id, template[TEXT])
+                    linkout[TEXT] = UNPREFIXED_GENE_REGEX.sub(
+                        unprefixed_id, linkout[TEXT]
+                    )
                     linkouts.append(linkout)
         return linkouts
 
@@ -98,3 +117,45 @@ def process_genomic_regions(self, ids):
                     linkout[TEXT] = GENOMIC_REGION_REGEX.sub(id, template[TEXT])
                     linkouts.append(linkout)
         return linkouts
+
+    def process_gene_families(self, ids):
+        linkouts = []
+        if self.linkout_lookup.get(GENE_FAMILY_LINKOUTS) is None:
+            return linkouts
+
+        type_lookup = self.linkout_lookup.get(GENE_FAMILY_LINKOUTS)
+
+        for id in ids:
+            prefix = ".".join(id.split(".")[0:1])
+            if type_lookup.get(prefix) is not None:
+                templates = type_lookup[prefix]
+                for template in templates:
+                    linkout = {}
+                    linkout[METHOD] = template[METHOD]
+                    # TODO: if method is POST, we probably need to do something with the
+                    # request body content
+                    linkout[HREF] = GENE_FAMILY_REGEX.sub(id, template[HREF])
+                    linkout[TEXT] = GENE_FAMILY_REGEX.sub(id, template[TEXT])
+                    linkouts.append(linkout)
+        return linkouts
+
+    def process_pan_gene_sets(self, ids):
+        linkouts = []
+        if self.linkout_lookup.get(PAN_GENE_SET_LINKOUTS) is None:
+            return linkouts
+
+        type_lookup = self.linkout_lookup.get(PAN_GENE_SET_LINKOUTS)
+
+        for id in ids:
+            prefix = ".".join(id.split(".")[0:2])
+            if type_lookup.get(prefix) is not None:
+                templates = type_lookup[prefix]
+                for template in templates:
+                    linkout = {}
+                    linkout[METHOD] = template[METHOD]
+                    # TODO: if method is POST, we probably need to do something with the
+                    # request body content
+                    linkout[HREF] = PAN_GENE_SET_REGEX.sub(id, template[HREF])
+                    linkout[TEXT] = PAN_GENE_SET_REGEX.sub(id, template[TEXT])
+                    linkouts.append(linkout)
+        return linkouts