Comparison of escape across strains¶

This notebook makes correlation plots of effects of mutations on escape from antibodies across strains of HIV Envelope.

First, import python modules:

In [1]:
import pandas as pd
from scipy import stats
import altair as alt
import numpy as np

# allow more rows for Altair
_ = alt.data_transformers.disable_max_rows()

Make plots:

In [2]:
for antibody in ['3BNC117', '10-1074']: 
    df_BF520 = pd.read_csv(f'../HIV_Envelope_BF520_DMS_3BNC117_10-1074/results/antibody_escape/averages/{antibody}_mut_effect.csv')
    df_TRO11 = pd.read_csv(f'results/antibody_escape/averages/{antibody}_mut_effect.csv')
    func_effects_BF520 = pd.read_csv(f'../HIV_Envelope_BF520_DMS_3BNC117_10-1074/results/func_effects/averages/TZM-bl_entry_func_effects.csv')
    func_effects_TRO11 = pd.read_csv(f'results/func_effects/averages/TZM-bl_entry_func_effects.csv')
    merged_df = df_BF520.merge(df_TRO11, on=['site', 'mutant'], suffixes=('_BF520', '_TRO11'))
    merged_df = merged_df.merge(func_effects_BF520[['site', 'mutant', 'effect']], on=['site', 'mutant'])
    merged_df = merged_df.merge(func_effects_TRO11[['site', 'mutant', 'effect']], on=['site', 'mutant'], suffixes=('_BF520', '_TRO11'))
    #merged_df = merged_df.query('site!="281"')
    for floored in [False, True]: 
        if floored: 
            plot_df = merged_df.copy()
            plot_df['escape_median_BF520'] = np.clip(plot_df['escape_median_BF520'], a_min=0, a_max=None)
            plot_df['escape_median_TRO11'] = np.clip(plot_df['escape_median_TRO11'], a_min=0, a_max=None)
        else: 
            plot_df = merged_df
        
        plot_df['max_std'] = plot_df[['escape_std_BF520', 'escape_std_TRO11']].values.max(1)
        plot_df['min_func_effect'] = plot_df[['effect_BF520', 'effect_TRO11']].values.min(1)
        plot_df = plot_df.assign(conserved=lambda x: x['wildtype_BF520']==x['wildtype_TRO11'])

        slope, intercept, r_value, p_value, std_err = stats.linregress(
        (plot_df
         .query('times_seen_BF520>3 & times_seen_TRO11>2')
         #.query('escape_std_BF520<1.5 & escape_std_TRO11<1.5')
         .query('effect_BF520>-4 & effect_TRO11>-4')
         ['escape_median_BF520']
         .astype(float)
        ),
        (plot_df
         .query('times_seen_BF520>3 & times_seen_TRO11>2')
         #.query('escape_std_BF520<1.5 & escape_std_TRO11<1.5')
         .query('effect_BF520>-4 & effect_TRO11>-4')
         ['escape_median_TRO11']
         .astype(float)
        ),
        )

        corr_chart = (
        alt.Chart((plot_df
                   .query('times_seen_BF520>3 & times_seen_TRO11>2')
                   #.query('escape_std_BF520<1.5 & escape_std_TRO11<1.5')
                   .query('effect_BF520>-4 & effect_TRO11>-4')
                  ))
            .encode(
                x=alt.X('escape_median_BF520',
                    #    scale=alt.Scale(type="log"),
                       ),
                y=alt.Y(
                    'escape_median_TRO11',
                    #scale=alt.Scale(type="log"),
                ),
                color=alt.Color("conserved"),
                tooltip=[
                    alt.Tooltip(c, format=".3g") if plot_df[c].dtype == float
                    else c
                    for c in plot_df.columns]
            )
        .mark_circle(filled=True, size=60, opacity=0.4)
        .configure_axis(grid=False)
        .resolve_scale(y="independent", x="independent")
        .properties(width=200, height=200)
        )
        print(f"R^2: {r_value**2}")
        print(f"R: {r_value}")
        corr_chart.display()

R^2: 0.058869542821923554
R: 0.24263046556836912

R^2: 0.02505079303290191
R: 0.15827442317981105

R^2: 0.29906373786543666
R: 0.5468672031356759

R^2: 0.4214133166189432
R: 0.6491635515175996
In [ ]: