dashboard.py

import warnings
warnings.filterwarnings("ignore")
import io
import os
import time
import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
warnings.simplefilter(action='ignore', category=RuntimeWarning)
import pandas as pd
import csv
import ast
from tqdm import tqdm
from operator import itemgetter
import numpy as np
import re
import datetime
import html
from joblib import Parallel, delayed
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
#plt.style.use('seaborn-paper')
import holoviews as hv
from holoviews import opts, dim
from bokeh.sampledata.les_mis import data
from bokeh.io import show
from bokeh.sampledata.les_mis import data
import panel as pn
#pn.extension(design='material')

import bokeh
from bokeh.resources import INLINE
from holoviews.operation.timeseries import rolling, rolling_outlier_std
hv.extension('bokeh')

## LOAD DATASETS

data = './data'


def read_freq_map(filename):
    df = pd.read_csv(os.path.join(data,filename), sep='	')
    #df = df.head(10)
    if 'Unnamed: 0' in df.columns:
        df = df.drop('Unnamed: 0', axis=1)
    column_0 = df.columns[0]
    column_1 = df.columns[1]
    freqmap = dict(zip(df[column_0], df[column_1]))
    return freqmap


def read_ont_freq_dataframe(filename):
    df = pd.read_csv(os.path.join(data,filename), sep='	')
    #print(df)
    if 'Unnamed: 0' in df.columns:
        df = df.drop('Unnamed: 0', axis=1)
    column_0 = df.columns[0]
    column_1 = df.columns[1]
    freqmap = dict(zip(df[column_0], df[column_1]))
    return freqmap


entityTypesFreqMap = read_freq_map('entityTypes.tsv')
relationTypesFreqMap = read_freq_map('relationTypes.tsv')
topDrugEntities = read_freq_map('topDrugs.tsv')
#print(topDrugEntities)
topConditionEntities = read_freq_map('topConditions.tsv')
topDrugOnts_df = pd.read_csv(os.path.join(data,'topDrugOntologies.tsv'), sep='\t')
topConditionOnts_df = pd.read_csv(os.path.join(data,'topConditionOntologies.tsv'), sep='\t')


grouping_filtered = pd.read_csv(os.path.join(data, 'drugReviewsCausal_relations.tsv'), sep="	")



################################# CREATE CHARTS ############################
def create_type_bar_charts(entRelsButton, **kwargs):
  if entRelsButton=='Entity':
    dictionary = entityTypesFreqMap
    return hv.Bars(dictionary, hv.Dimension('Entity Types'), 'Frequency').opts( framewise=True, xrotation=45,width=1200, height=600)
  elif entRelsButton=='Relation':
    dictionary = relationTypesFreqMap
    return hv.Bars(dictionary, hv.Dimension('Relation Types'), 'Frequency').opts(framewise=True, xrotation=45,width=1200, height=600)


def create_ent_bar_charts(ents, **kwargs):
# Create button widgets for each label
    drug_buttons = []
    condition_buttons = []
    for i, drg in enumerate(list(topDrugEntities.keys())):
        button = pn.widgets.Button(name=drg, width=150)
        ## Open the associated URL in a new tab when button is clicked
        button.js_on_click(code=f'window.open("https://api-vast.jrc.service.ec.europa.eu/describe/?url=http://causaldrugskg.org/causaldrugskg/resource/{drg}", "_blank");')
        drug_buttons.append(button)
    for i, cnd in enumerate(list(topConditionEntities.keys())):
        button = pn.widgets.Button(name=cnd, width=150)
        ## Open the associated URL in a new tab when button is clicked
        button.js_on_click(
            code=f'window.open("https://api-vast.jrc.service.ec.europa.eu/describe/?url=http://causaldrugskg.org/causaldrugskg/resource/{cnd}", "_blank");')
        condition_buttons.append(button)

        # Stack the buttons vertically (or wrap in a GridBox for nicer layout)
    drug_button_column = pn.Column(*drug_buttons, sizing_mode='stretch_width')
    condition_button_column = pn.Column(*condition_buttons, sizing_mode='stretch_width')


    if ents=='Drug':
        dictionary = topDrugEntities
        bars = hv.Bars(dictionary, hv.Dimension('Drug Entities'), 'Frequency').opts(framewise=True, xrotation=45,width=1200, height=600,  fontsize={'xticks': 18, 'xlabel':18, 'ylabel':16})
        # Combine everything into a Panel layout
        layout = pn.Row(bars, drug_button_column)
        return layout
    elif ents=='Condition':
        dictionary = topConditionEntities
        bars = hv.Bars(dictionary, hv.Dimension('Condition Entities'), 'Frequency').opts(framewise=True, xrotation=45,width=1200, height=600, fontsize={'xticks': 18, 'xlabel':18, 'ylabel':16})
        layout = pn.Row(bars, condition_button_column)
        return layout


def create_ontology_bar_charts(ents, **kwargs):
    if ents=='Drug':
        df = pd.DataFrame({
            'Drug_Ontologies': [ont.split('/')[-1] for ont in topDrugOnts_df['ontology']],
            'Frequency': list(topDrugOnts_df['count']),
            'url': list(topDrugOnts_df['ontology_url'])  # using full keys as hyperlinks
        })
        drug_ontolgy_buttons = []
        for i,row in df.iterrows():
            button = pn.widgets.Button(name=row['Drug_Ontologies'], width=150)
            ## Open the associated URL in a new tab when button is clicked
            url = row["url"]
            button.js_on_click(
                code=f'window.open("{url}", "_blank");')
            drug_ontolgy_buttons.append(button)
        drug_ontology_column = pn.Column(*drug_ontolgy_buttons, sizing_mode='stretch_width')

        # Create bar chart with label as x-axis
        bars = hv.Bars(df, kdims=['Drug_Ontologies'], vdims=['Frequency'])
        bars.opts(
            framewise=True,
            tools=['hover'],
            width=1200,
            height=600,
            show_legend=True,
            xrotation=45,
            xlabel='Drug_Ontologies',
            ylabel='Frequency',
            hover_tooltips=[
                ("Drug_Ontologies", "@Drug_Ontologies"),
                ("Frequency", "@Frequency")
            ]
        )
        #links_panel = pn.Column(*[pn.pane.Markdown(f"[{row.Drug_Ontologies}]({row.url})", width=400) for _, row in df.iterrows()],name='Links')
        layout = pn.Row(bars, drug_ontology_column)
        return layout
    elif ents=='Condition':
        df = pd.DataFrame({
              'Condition_Ontologies': [ont.split('/')[-1] for ont in topConditionOnts_df['ontology']],
              'Frequency': list(topConditionOnts_df['count']),
              'url': list(topConditionOnts_df['ontology_url'])  # using full keys as hyperlinks
          })
        condition_ontolgy_buttons = []
        for i, row in df.iterrows():
            button = pn.widgets.Button(name=row['Condition_Ontologies'], width=150)
            ## Open the associated URL in a new tab when button is clicked
            url = row["url"]
            button.js_on_click(
                code=f'window.open("{url}", "_blank");')
            condition_ontolgy_buttons.append(button)
        condition_ontology_column = pn.Column(*condition_ontolgy_buttons, sizing_mode='stretch_width')
        # Create bar chart with label as x-axis
        bars = hv.Bars(df, kdims=['Condition_Ontologies'], vdims=['Frequency'])
        bars.opts(
        framewise=True,
        tools=['hover'],
        width=1200,
        height=600,
        show_legend=True,
        xrotation=45,
        xlabel='Condition_Ontologies',
        ylabel='Frequency',
        hover_tooltips=[
            ("Condition Ontologies", "@Condition Ontologies"),
            ("Frequency", "@Frequency") ])
        #links_panel = pn.Column(*[pn.pane.Markdown(f"[{row.Condition_Ontologies}]({row.url})", width=400) for _, row in df.iterrows()],name='Links')
        layout = pn.Row(bars, condition_ontology_column)
        return layout

############################# WIDGETS & CALLBACK ###########################################

def filter_data0(df, min_value):
    filtered_df = df[df['value'] >= min_value]
    return filtered_df


def plot_chord(df,min_value):
    filtered_df = filter_data0(df, min_value)
  # Create a Holoviews Dataset for nodes
    nodes = hv.Dataset(filtered_df, 'index')
    nodes.data.head()
    chord = hv.Chord(filtered_df, ['source', 'target'], ['value'])
    return chord.opts(opts.Chord(cmap='Category20', edge_cmap='Category20', label_text_color="white",  node_color = hv.dim('index').str(),  edge_color = hv.dim('source').str(), labels = 'index', tools=['hover'],   width=800, height=800))


def chordify_triples(rel_grouping, min_val):
     # Define range for minimum value slider
    min_value_range = rel_grouping['value'].unique()
    min_value_range.sort()
    min_value_range = min_value_range[min_value_range > min_val]


    # Define HoloMap with minimum value and attribute as key dimensions
    holomap = hv.HoloMap({min_value: plot_chord(rel_grouping, min_value)
                          for min_value in min_value_range},
                         kdims=['Show triples with support greater than']
                         )
    return holomap


# https://tabler-icons.io/
button1 = pn.widgets.Button(name="Introduction", button_type="warning", icon="file-info", styles={"width": "100%"})
button2 = pn.widgets.Button(name="Top Key Entities", button_type="warning", icon="chart-bar", styles={"width": "100%"})
button3 = pn.widgets.Button(name="Entity/Relation Types:", button_type="warning",  icon="chart-histogram", styles={"width": "100%"})
button4 = pn.widgets.Button(name="Ontology Coverage", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})
#button5 = pn.widgets.Button(name="Causal Relation Chord Diagrams", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})

markdown_button_style = """
<div style="background-color: #f0f0f0; /* Matches 'warning' button type */
    color: white;
    font-weight: bold;
    padding: 8px 12px;
    border-radius: 4px;
    text-align: center;
    width: 100%;
">
Causal Relation Chord Diagrams
</div>
"""
#button5 = pn.pane.Markdown(markdown_button_style, width_policy="max")

button5 = pn.pane.Markdown("<div style='background-color:#f7c045; color: black; padding:8px 12px; font-weight: bold; border:1px solid #ccc; " "border-radius:8px; text-align:center; width:100%; white-space: nowrap;'>Causal Relation Chord Diagrams</div>", width=225, height=30, margin=(9, 9))


# Define child buttons
child_button_1 = pn.widgets.Button(name="Cause", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})
child_button_2 = pn.widgets.Button(name="Enable", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})
child_button_3 = pn.widgets.Button(name="Prevent", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})
child_button_4 = pn.widgets.Button(name="Hinder", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})
child_button_5 = pn.widgets.Button(name="Other", button_type="warning", icon="chart-dots-filled", styles={"width": "100%"})
# Layout: dendrogram-style using vertical + indent
tree_layout = pn.Column(
    button5,
    pn.Row(pn.Spacer(width=70),  # indent
           pn.Column(child_button_1, child_button_2,child_button_3,child_button_4, child_button_5))
)

entRelsButton = pn.widgets.RadioButtonGroup(name='### Select', options=['Entity','Relation'], value = 'Entity' )

entTypeButton = pn.widgets.RadioButtonGroup(name='### Select Entity Type', options=list(entityTypesFreqMap.keys()), value='Drug')

#relationTypeButton = pn.widgets.RadioButtonGroup(options=list(relationTypesFreqMap.keys()), value='Cause', name='Select Causal Relation')

# Define the callback function to update the HoloMap
#def update_holomap(event):
#    initial_holomap.object = filter_triples(event.new)


# Create the initial HoloMap
#initial_holomap = filter_triples(relationTypeButton.value)

# Bind the callback function to the value change event of the RadioButton widget
#relationTypeButton.param.watch(update_holomap, 'value')


def show_page(page_key):
    main_area.clear()
    main_area.append(mapping[page_key])

button1.on_click(lambda event: show_page("Page1"))
button2.on_click(lambda event: show_page("Page2"))
button3.on_click(lambda event: show_page("Page3"))
button4.on_click(lambda event: show_page("Page4"))
#button5.on_click(lambda event: show_page("Page5"))
child_button_1.on_click(lambda event: show_page("Page5a"))
child_button_2.on_click(lambda event: show_page("Page5b"))
child_button_3.on_click(lambda event: show_page("Page5c"))
child_button_4.on_click(lambda event: show_page("Page5d"))
child_button_5.on_click(lambda event: show_page("Page5e"))


### CREATE PAGE LAYOUTS

def CreatePage1():
    return pn.Column(pn.pane.Markdown("""

This is a dashboard for exploring a causal relation knowledge graph automatically extracted from a collection of drug reviews. The source data consists of around 19200 reviews from the **Drug Reviews (Druglib.com)**  dataset (https://archive.ics.uci.edu/dataset/461/drug+review+dataset+druglib+com) containing patient reviews on specific drugs along with related conditions, crawled from online pharmaceutical review sites.
The causal relations represented in the KG are defined by the **MIMICause** schema (https://huggingface.co/datasets/pensieves/mimicause). The underlying CausalDrugsKG graph is available in Turtle and RDF serialization format in the European Data portal: https://data.jrc.ec.europa.eu/dataset/acebeb4e-9789-4b5c-97ec-292ce14e75d0.


---------------------------

## Top Key Entities
Bar plots representing the occurence counts of the top 30 Drug and Condition entities in the KG, where occurrence means the entity is either the Subject or Object of an extracted triple in the KG.
Clicking on the entity name in the right legend redirects to the corresponding entry in the Virtuoso Faceted Browser endpoint of the KG 


## Entities/Relation Types 
Bar plots of the Entity and Relation type counts.


## Ontology Coverage
Bar plots representing the linking of KG entities to standard Biomedical ontologies. Bar heights indicate the number of Drug/Condition entities linked to the corresponding ontology.
Linking is performed using the Bioportal API (https://bioportal.bioontology.org/)
Clicking on the ontology name on the right legend redirects to the ontology entry page.


## Causal Relations Chord Diagrams
Entity Chord Diagrams represent the most frequently connected entity pairs within the KG through chord illustrations, serving as both Subjects and Objects of predicative triples. The size of the chords corresponds to the support of the depicted relations.
""", width=800), align="center")



def CreatePage2():
    return pn.Column(
        pn.pane.Markdown("## Top 30 Entities "),
        entTypeButton,
        pn.bind(create_ent_bar_charts, entTypeButton),
        align="center", )


def CreatePage3():
    return pn.Column(
        pn.pane.Markdown("## Entity/Relation Types "),
        entRelsButton,
        pn.bind(create_type_bar_charts, entRelsButton),
        align="center",
    )

def CreatePage4():
    return pn.Column(
        pn.pane.Markdown("## Bio-Medical Ontology Coverage "),
        entTypeButton,
        pn.bind(create_ontology_bar_charts, entTypeButton),
        align="center", )
def CreatePage5():
    return pn.Column(
        pn.pane.Markdown("## Causal Relation Chord Diagrams"),
        chordify_triples(grouping_filtered),
        align="center", )

def CreatePage5a():
    rel_grouping = grouping_filtered[grouping_filtered['causal_relation'] == 'Cause']
    return pn.Column(
        pn.pane.Markdown("## Relation Chord Diagram: Cause"),
        chordify_triples(rel_grouping,10),
        align="center", )

def CreatePage5b():
    rel_grouping = grouping_filtered[grouping_filtered['causal_relation'] == 'Enable']
    return pn.Column(
        pn.pane.Markdown("## Relation Chord Diagram: Enable"),
        chordify_triples(rel_grouping,4),
        align="center", )

def CreatePage5c():
    rel_grouping = grouping_filtered[grouping_filtered['causal_relation'] == 'Prevent']
    return pn.Column(
        pn.pane.Markdown("## Relation Chord Diagram: Prevent"),
        chordify_triples(rel_grouping,50),
        align="center", )

def CreatePage5d():
    rel_grouping = grouping_filtered[grouping_filtered['causal_relation'] == 'Hinder']
    return pn.Column(
        pn.pane.Markdown("## Relation Chord Diagram: Hinder"),
        chordify_triples(rel_grouping,10),
        align="center", )

def CreatePage5e():
    rel_grouping = grouping_filtered[grouping_filtered['causal_relation'] == 'Other']
    return pn.Column(
        pn.pane.Markdown("## Relation Chord Diagram: Other"),
        chordify_triples(rel_grouping,10),
        align="center", )
mapping = {
    "Page1": CreatePage1(),
    "Page2": CreatePage2(),
    "Page3": CreatePage3(),
    "Page4": CreatePage4(),
    #"Page5": CreatePage5(),
    "Page5a": CreatePage5a(),
    "Page5b": CreatePage5b(),
    "Page5c": CreatePage5c(),
    "Page5d": CreatePage5d(),
    "Page5e": CreatePage5e(),
}

#################### SIDEBAR LAYOUT ##########################
sidebar = pn.Column(pn.pane.Markdown("## Panels"), button1,button2,button3,
                    button4,tree_layout,
                   styles={"width": "100%", "padding": "15px"})

#################### MAIN AREA LAYOUT ##########################
main_area = pn.Column(mapping["Page1"], styles={"width":"100%"})

###################### APP LAYOUT ##############################
template = pn.template.BootstrapTemplate(
    title=" CausalDrugsKG_Dashboard ",
    sidebar=[sidebar],
    main=[main_area],
    #header_background="black",
    #site="Charting the Landscape of Digital Health",
    #theme=pn.template.DarkTheme,
    sidebar_width=270, ## Default is 330
    busy_indicator=pn.indicators.BooleanStatus(value=True),
)

### DEPLOY APP

# Serve the Panel app
template.servable()