Merge pull request #110 from graphistry/bio-orth

Bio orth

src/python/requirements-app.txt

@@ -1,3 +1,3 @@
 ###
 ### Use this file for your custom view requirements here vs in the requirements-system.txt
-###
+###

src/python/views/demo_bio_01_funcoup/__init__.py

@@ -0,0 +1,272 @@
+import graphistry, os, pandas as pd, streamlit as st
+from components import GraphistrySt, URLParam
+from graphistry import PyGraphistry
+from css import all_css
+from time import sleep
+import logging
+
+############################################
+#
+#   DASHBOARD SETTINGS
+#
+############################################
+#  Controls how entrypoint.py picks it up
+
+
+app_id = 'app_bio_01'
+logger = logging.getLogger(app_id)
+urlParams = URLParam(app_id)
+
+
+def info():
+    return {
+        'id': app_id,
+        'name': 'Bio: FUNCOUP',
+        'enabled': True,
+        'tags': ['bio', 'large', 'funcoup','demo'],
+    }
+
+
+def run():
+    run_all()
+
+
+############################################
+#
+#   CUSTOM CSS
+#
+############################################
+# Have fun!
+
+def custom_css():
+
+    all_css()  # our favorites
+
+
+############################################
+#
+#   SIDEBAR RENDER AERA
+#
+############################################
+# Given URL params, render left sidebar form and return combined filter settings
+
+# #https://docs.streamlit.io/en/stable/api.html#display-interactive-widgets
+def sidebar_area():
+    st.sidebar.title("Select a Species")
+    species_to_label = {
+        'A.thaliana': 'A.thaliana',
+        'B.subtilis': 'B.subtilis',
+        'B.taurus': 'B.taurus',
+        'C.elegans': 'C.elegans',
+        'C.familiaris': 'C.familiaris',
+        'C.intestinalis': 'C.intestinalis',
+        'D.discoideum': 'D.discoideum',
+        'D.melanogaster': 'D.melanogaster',
+        'D.rerio': 'D.rerio',
+        'E.coli': 'E.coli',
+        'G.gallus': 'G.gallus',
+        'H.sapiens': 'H.sapiens',
+        'M.jannaschii': 'M.jannaschii',
+        'M.musculus': 'M.musculus',
+        'O.sativa': 'O.sativa',
+        'P.falciparum': 'P.falciparum',
+        'R.norvegicus': 'R.norvegicus',
+        'S.cerevisiae': 'S.cerevisiae',
+        'S.pombe': 'S.pombe',
+        'S.scrofa': 'S.scrofa',
+        'S.solfataricus': 'S.solfataricus',
+    }
+
+    base_url = os.environ['BASE_URL']
+
+    filter_by_org_type_init = urlParams.get_field('filter_by_org', default='B.subtilis')
+    filter_by_org_type = \
+        st.sidebar.selectbox(
+            'Choose organism:',
+            ('A.thaliana', 'B.subtilis', 'B.taurus','C.elegans','C.familiaris','C.intestinalis','D.discoideum','D.melanogaster','D.rerio','E.coli','G.gallus','H.sapiens','M.jannaschii','M.musculus','O.sativa','P.falciparum','R.norvegicus','S.cerevisiae','S.pombe','S.scrofa','S.solfataricus'),
+            index=('A.thaliana', 'B.subtilis', 'B.taurus','C.elegans','C.familiaris','C.intestinalis','D.discoideum','D.melanogaster','D.rerio','E.coli','G.gallus','H.sapiens','M.jannaschii','M.musculus','O.sativa','P.falciparum','R.norvegicus','S.cerevisiae','S.pombe','S.scrofa','S.solfataricus').index(filter_by_org_type_init),
+            format_func=(lambda option: species_to_label[option]))
+    urlParams.set_field('filter_by_org', filter_by_org_type)
+
+    st.sidebar.title("Select a Network Type")
+    umap_to_label = {
+        True: 'UMAP',
+        False: 'FunCoup',
+    }
+
+    filter_by_umap_type_init = urlParams.get_field('filter_by_umap', default='FunCoup')
+    filter_by_umap_type = \
+        st.sidebar.selectbox(
+            'Display functional coupling network (select link evidence below) or UMAP against all 40 evidence types:',
+            (True,False),
+            index=(True,False).index(filter_by_umap_type_init),
+            format_func=(lambda option: umap_to_label[option]))
+    urlParams.set_field('filter_by_umap', filter_by_umap_type)
+
+    if filter_by_umap_type is 'UMAP':
+        filter_by_net_type = 'full'
+    else:
+        filter_by_net_type = 'compact'
+
+    st.sidebar.title("Select an Evidence Type")
+    edge_to_label = {'PFC':'PFC', 'FBS_max':'FBS_max'}
+
+    filter_by_node_type_init = urlParams.get_field('filter_by_node', default='PFC')
+    filter_by_node_type = \
+        st.sidebar.selectbox(
+            'for FunCoup Network display',
+            ('PFC', 'FBS_max'),
+            index=('PFC', 'FBS_max').index(filter_by_node_type_init),
+            format_func=(lambda option: edge_to_label[option]))
+    urlParams.set_field('filter_by_node', filter_by_node_type)
+
+
+    edges_df = pd.read_csv('https://funcoup.org/downloads/download.action?type=network&instanceID=24480085&fileName=FC5.0_'+filter_by_org_type+'_'+filter_by_net_type+'.gz', sep='\t')
+
+    return {
+        'edges_df': edges_df,
+        'node_type': filter_by_node_type,
+        'umap_type': filter_by_umap_type,
+    }
+
+
+############################################
+#
+#   FILTER PIPELINE
+#
+############################################
+# Given filter settings, generate/cache/return dataframes & viz
+
+#@st.cache(suppress_st_warning=True, allow_output_mutation=True)
+@st.cache_data
+def run_filters(edges_df, node_type, umap_type=False):
+
+    filtered_edges_df = edges_df
+    # filtered_edges_df = filtered_edges_df.replace({'ENSG00000':''},regex=True)
+    filtered_edges_df.columns=filtered_edges_df.columns.str.split(':').str[1]
+
+    # include viz generation as part of cache
+    url = plot_url(filtered_edges_df,node_type,umap_type)
+
+    return {
+        'edges_df': filtered_edges_df,
+        'url': url,
+    }
+
+
+############################################
+#
+#   VIZ
+#
+############################################
+
+
+def plot_url(edges_df,node_type, umap_type=False):
+
+    edges_df.replace({'ENSG00000':''},regex=True,inplace=True) ## remove ENSG00000 from gene names for better compression
+
+    nodes_df = pd.DataFrame({
+        'n': pd.concat([edges_df['Gene1'], edges_df['Gene2']]).unique()
+    })
+    n = len(nodes_df)
+
+    nodes_df['ind'] = nodes_df.index 
+    nodes_df['nc'] = nodes_df['ind'].apply(lambda v: 0x01000000 * round(255 * v / n,2))
+
+    logger.info('Starting graphistry plot')
+    if not GraphistrySt().test_login():
+        return ''
+
+    if umap_type == False:
+        url = graphistry\
+                .edges(edges_df)\
+                .bind(source="Gene1", destination="Gene2", edge_weight=node_type)\
+                .nodes(nodes_df)\
+                .bind(node='n', point_color='nc')\
+                .settings(url_params={
+                    'pointSize': 0.3,
+                    'splashAfter': 'false',
+                    'bg': '%23' + 'f0f2f6'
+                })\
+                .plot(render=False)#, as_files=True, suffix='.html', output=None, open=False)
+    elif umap_type == True:
+
+        AA = graphistry\
+                .nodes(edges_df)\
+                .bind(source="Gene1", destination="Gene2")\
+                .settings(url_params={
+                    'pointSize': 0.3,
+                    'splashAfter': 'false',
+                    'bg': '%23' + 'f0f2f6'
+                })\
+                .umap(feature_engine='dirty_cat',engine='umap_learn',memoize=True)
+        emb2=AA._node_embedding
+        url=graphistry.nodes(emb2.reset_index(),'index').edges(AA._edges,'_src_implicit','_dst_implicit').bind(point_x="x",point_y="y").settings(url_params={"play":0}).addStyle(bg={'color': '#eee'}).plot(render=False)
+
+    logger.info('Generated viz, got back urL: %s', url)
+
+    return url
+
+
+############################################
+#
+#   MAIN RENDER AERA
+#
+############################################
+# Given configured filters and computed results (cached), render
+
+def main_area(edges_df, url):
+
+    logger.debug('rendering main area, with url: %s', url)
+    gst = GraphistrySt()
+    if PyGraphistry._is_authenticated:
+        gst.render_url(url)
+    else:
+        st.title("Welcome to graph-app-kit!")
+        st.write("""
+            This particular demo requires configuring your graph-app-kit with service credentials for
+            accessing your Graphistry server
+
+            If this is the first time you are seeing graph-app-kit, it is Graphistry's open-source extension
+            of the https://streamlit.io/ low-code Python dashboarding tool. It adds:
+              * Optional Docker, Docker Compose, and AWS CloudFormation self-hosted quick launchers
+              * Multiple dashboard support
+              * Optional GPU & AI dependencies (Nvidia driver, RAPIDS, PyTorch) aligned with Graphistry releases
+              * Graph computing dependencies (Gremlin, TigerGraph, ...)
+              * A Graphistry plotting component
+
+            Starting with Graphistry 2.39, graph-app-kit comes prebundled:
+              * Public and staff-only Private dashboards
+              * Control access via User -> Admin port -> DJANGO-WAFFLE -> Flags"
+              * ... then edit to desired visibility for flag_show_public_dashboard, flag_show_private_dashboard
+              * ... and optionally prevent running of the services via your docker-compose.override.yml
+        """)
+
+
+############################################
+#
+#   Putting it all together
+#
+############################################
+
+def run_all():
+
+    custom_css()
+
+    try:
+
+        # Render sidebar and get current settings
+        sidebar_filters = sidebar_area()
+
+        # logger.debug('sidebar_filters: %s', sidebar_filters)
+
+        # Compute filter pipeline (with auto-caching based on filter setting inputs)
+        # Selective mark these as URL params as well
+        filter_pipeline_result = run_filters(**sidebar_filters)
+
+        # Render main viz area based on computed filter pipeline results and sidebar settings
+        main_area(**filter_pipeline_result)
+
+    except Exception as exn:
+        st.write('Error loading dashboard')
+        st.write(exn)