dashboard.py

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import streamlit as st
import missingno as msno
import matplotlib.patches as mpatches
import numpy as np
from sklearn.preprocessing import StandardScaler
#from sklearn.cluster import KMeans
#from sklearn.preprocessing import StandardScaler
#from sklearn.decomposition import PCA

#https://streamlit.io

# Load spreadsheet with studies 
file_path = "ModelCat_paper__020624.xlsx"
df = pd.read_excel(file_path, sheet_name='Sheet1')
df['Sex'] = pd.concat([df['Sex1'], df['Sex2']]).reset_index(drop=True)

st.title("PRECISE-TBI Metadata Dashboard")

# Basic Summary
st.subheader("Metadata Summary")
categorical_columns =  ["Sex", "Species", "Strain Type", "TBI Model Type", "metadata:age:category"]



st.subheader("Summary Preview")
st.dataframe(df.head())

st.write({
    "Total Rows": len(df),
    "Total Columns": len(df.columns),
    "Missing Values": df.isnull().sum().sum(),
    "Columns with Missing Values": df.isnull().sum()[df.isnull().sum() > 0].to_dict(),
    "Column Names": df.columns
})
    

for col in categorical_columns:
    if col in df.columns:
        st.subheader(f"Distribution of {col}")
        fig, ax = plt.subplots()
        sns.countplot(data=df, y="TBI Model Type", hue=col, ax=ax)
        st.pyplot(fig)

required_columns = ["Species", "Strain Type"]

if not all(col in df.columns for col in required_columns):
    st.error(required_columns)
else:
    # Drop rows with missing values in the required columns
    df = df.dropna(subset=required_columns)

    # Get unique species
    species_list = df["Species"].unique()

    # Iterate over each species and plot the distribution of strains
    for species in species_list:
        st.subheader(f"Distribution of strains for species: {species}")
        species_df = df[df["Species"] == species]
        fig, ax = plt.subplots()
        sns.countplot(y=species_df["Strain Type"], order=species_df["Strain Type"].value_counts().index, ax=ax)
        ax.set_title(f"Strain Distribution for {species}")
        st.pyplot(fig)

# General Summary
st.subheader("General Summary")
df=df.dropna(subset=['TBI Model Type'])


df_filt = df[["Sex", "Species", "Strain", "TBI Model","Age category", "Age (weeks)", "Weight (grams)", "Device Name"]]
df_filt.replace(r'^\s*$', np.nan, regex=True, inplace=True)
fig,ax = plt.subplots(figsize=(10,5))
msno.matrix(df_filt, ax=ax, fontsize=12, color= (0.93, 0.00, 0.37), sparkline=False)

red_patch = mpatches.Patch(color= (0.93, 0.00, 0.37), label='Data present')
white_patch = mpatches.Patch(color='white', label='Data absent')

ax.legend(handles=[red_patch, white_patch],loc='center left', bbox_to_anchor=(1.2, 0.7))

st.pyplot(fig)


# Age Classification Analysis

df['Age_min(weeks)'] = pd.to_numeric(df['Age_min(weeks)'], errors='coerce')
df['Weight_min(grams)'] = pd.to_numeric(df['Weight_min(grams)'], errors='coerce')

df_clean = df.dropna(subset=['Age_min(weeks)', 'Weight_min(grams)'])

df['Sex'] = df['Sex'].astype('category')
df['Species'] = df['Species'].astype('category')

st.subheader("Age and Weight Distribution by Sex and Species")
fig, ax = plt.subplots(figsize=(10, 6))
sns.scatterplot(data=df, x='Age_min(weeks)', y='Weight_min(grams)', hue='Sex', style='Species', ax=ax)
ax.set_xlabel('Age (weeks)')
ax.set_ylabel('Weight (grams)')
st.pyplot(fig)

# Strain Analysis
st.subheader("Strain Classification Analysis")
df['strain_prefix'] = df['Strain'].str[:4]
strain_summary = df.groupby(['strain_prefix', 'Strain']).size().reset_index(name='Count')
fig, ax = plt.subplots(figsize=(10, 6))
sns.countplot(data=strain_summary, y='strain_prefix', hue='Strain', ax=ax)
ax.set_ylabel('Strain Prefix')
ax.set_xlabel('Count')
st.pyplot(fig)

# Model Summary
st.subheader("Model Summary by Age, Weight, Classification, Species, and Strain")

model_summary = df.groupby(['TBI Model Type', 'Species']).agg({
'Age_min(weeks)': lambda x: pd.to_numeric(x, errors='coerce').mean(),
'Age_max(weeks)': lambda x: pd.to_numeric(x, errors='coerce').mean(),
'Weight_min(grams)': lambda x: pd.to_numeric(x, errors='coerce').mean(),
'Weight_max(grams)': lambda x: pd.to_numeric(x, errors='coerce').mean(),
'Sex': pd.Series.nunique,
'Strain': pd.Series.nunique
}).reset_index()


st.write(model_summary)



#Missing data analysis - CCI
st.subheader("CCI Model Papers - Missing Data Summary")
cci_col = ["TBI Model Type", "TBI Model", "metadata:tbi_device:type", "Device Name", "Impact Depth (mm)_min","Impact Depth (mm)_max","Impact Duration (ms)-min","Impact Duration (ms)-max","Impact Velocity (m/s)_min","Impact Velocity (m/s)_max","Impactor Tip_min (mm)","Impactor Tip_max(mm)","Impactor Shape","metadata:tbi_device:company", "metadata:tbi_device:company_geo"]

cci_newdf = df[cci_col]

cci_df = cci_newdf[cci_newdf["TBI Model Type"]== "Controlled cortical impact model"]

fig,ax = plt.subplots(figsize=(10,5))
msno.matrix(cci_df, ax=ax, fontsize=12, color= (0.93, 0.00, 0.37), sparkline=False)

red_patch = mpatches.Patch(color= (0.93, 0.00, 0.37), label='Data present')
white_patch = mpatches.Patch(color='white', label='Data absent')

ax.legend(handles=[red_patch, white_patch],loc='center left', bbox_to_anchor=(1.2, 0.7))

st.pyplot(fig)


# correlations?

numeric_cols = ['Age_min(weeks)', 'Age_max(weeks)', 'Weight (grams)',
       'Weight_min(grams)', 'Weight_max(grams)','Impact Depth (mm)_min',
       'Impact Depth (mm)_max', 'Impact Duration (ms)-min',
       'Impact Duration (ms)-max', 'Impact Velocity (m/s)_min',
       'Impact Velocity (m/s)_max', 'Impactor Tip_min (mm)',
       'Impactor Tip_max(mm)']

fig,ax = plt.subplots(figsize=(10,5))
msno.heatmap(df)
st.pyplot(fig)

# count the occurrences of each unique value 
species_counts = filtered_species["Species"].value_counts()


st.write("Count of different species variations:")
st.write(species_counts)

# bar chart to  compare
st.write("Bar Chart of Species Variations:")
st.bar_chart(species_counts)