fig1.py

import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
from matplotlib.gridspec import GridSpec
from util import *
import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd
from matplotlib.gridspec import GridSpec

data_source_file = './CSPRANK.csv'
parsed_data = pd.read_csv(data_source_file)

# parsed_data['AF2_TM_top_rank'] = pd.to_numeric(parsed_data['AF2_TM_top_rank'], errors='coerce')
# parsed_data['AF2_TM_top_rank'] = parsed_data['AF2_TM_top_rank'].astype(float)
parsed_data.dropna(subset=['AF2_TM_top_rank','AF2_DockQ_top_rank'], inplace=True)

# Ensure columns are numeric
parsed_data['AF2_TM_top_rank'] = pd.to_numeric(parsed_data['AF2_TM_top_rank'], errors='coerce')
parsed_data['AF2_DockQ_top_rank'] = pd.to_numeric(parsed_data['AF2_DockQ_top_rank'], errors='coerce')

# Drop rows that are still NaN
parsed_data.dropna(subset=['AF2_TM_top_rank','AF2_DockQ_top_rank'], inplace=True)

fig = plt.figure(figsize=(10, 10))
gs = GridSpec(4, 4, figure=fig)

ax_main = fig.add_subplot(gs[1:4, 0:3])
ax_x_hist = fig.add_subplot(gs[0, 0:3], sharex=ax_main)
ax_y_hist = fig.add_subplot(gs[1:4, 3], sharey=ax_main)

sns.scatterplot(
    x='AF2_DockQ_top_rank', 
    y='AF2_TM_top_rank', 
    data=parsed_data, 
    ax=ax_main
)
ax_main.set_xlabel('AF2 DockQ Top Rank')
ax_main.set_ylabel('AF2 TM Top Rank')
ax_main.axhline(y=0.8, color='red', linestyle='--')
ax_main.axvline(x=0.4, color='red', linestyle='--')
ax_main.fill_betweenx(y=[0.8, 1], x1=0.4, x2=1, color='gray', alpha=0.3)
ax_main.set_title('Scatterplot: AF2 DockQ Top Rank vs AF2 TM Top Rank')
ax_main.set_xlim(0, 1)
ax_main.set_ylim(0, 1)

sns.histplot(
    data=parsed_data, 
    x='AF2_DockQ_top_rank', 
    bins=30, 
    kde=False, 
    ax=ax_x_hist
)
ax_x_hist.set_ylabel('Frequency')
ax_x_hist.set_xlim(0, 1)
# ax_x_hist.set_title('Histogram: AF2 DockQ Top Rank')

# print(parsed_data['AF2_TM_top_rank'])
# parsed_data['AF2_TM_top_rank'] = pd.to_numeric(parsed_data['AF2_TM_top_rank'], errors='coerce')
# print(parsed_data['AF2_TM_top_rank'])

sns.histplot(
    data=parsed_data, 
    y='AF2_TM_top_rank', 
    bins=30, 
    kde=False, 
    ax=ax_y_hist, 
    orientation='horizontal'
)
ax_y_hist.set_xlabel('Frequency')
ax_y_hist.set_ylim(0, 1)
ax_y_hist.set_title('Histogram: AF2 TM Top Rank')

# Hide overlapping tick labels
plt.setp(ax_x_hist.get_xticklabels(), visible=False)
plt.setp(ax_y_hist.get_yticklabels(), visible=False)

plt.tight_layout()
plt.savefig('./Figures/AF2_DockQ_vs_TM_top_rank_scatterplot_with_histograms.png')
plt.close()