import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from cypsummary import *
import glob, re, sys, os, importlib
import cypsummary
importlib.reload(cypsummary)
del analyze
from cypsummary import *

/net/galaxy/home/koes/dkoes/.local/lib/python3.10/site-packages/matplotlib/projections/__init__.py:63: UserWarning: Unable to import Axes3D. This may be due to multiple versions of Matplotlib being installed (e.g. as a system package and as a pip package). As a result, the 3D projection is not available.
  warnings.warn("Unable to import Axes3D. This may be due to multiple versions of "

df = pd.concat((pd.read_csv(f,index_col=0) for f in glob.glob('*.csv')), ignore_index=True)

df["ASP_min"] = df[[
    "ASP311_OD1_ave",
    "ASP311_OD2_ave",
    "ASP304_OD1_ave",
    "ASP304_OD2_ave"
]].min(axis=1)

df

prefdf = df[~(df.featom.str.startswith('O') | df.featom.str.startswith('CL'))]
prefdf

df_min = prefdf.loc[
    prefdf.groupby(['cyp', 'ligand', 'charged'])['avefe'].idxmin()
].reset_index(drop=True)

def plot_scatter(df_max):
    fig, ax = plt.subplots(figsize=(8, 6))

    df_max["ASP_min"] = df_max[[
        "ASP311_OD1_ave",
        "ASP311_OD2_ave",
        "ASP304_OD1_ave",
        "ASP304_OD2_ave"
    ]].min(axis=1)

    # Base scatter (color and shape by ligand/cyp)
    sns.scatterplot(
        data=df_max,
        x="ASP_min",
        y="avefe",
        hue="ligand",
        style="cyp",
        s=80,            # marker size
        edgecolor="black",  # temporary edge
        linewidth=1,
        ax=ax
    )
    
    # Overlay edges by 'charged' state
    # Suppose charged is boolean or categorical
    for charge_state, subdf in df_max.groupby("charged"):
        sns.scatterplot(
            data=subdf,
            x="ASP_min",
            y="avefe",
            hue=None,
            style="cyp",
            markers=True,
            s=80,
            facecolor="none",
            edgecolor="black" if charge_state else "white",
            linewidth=1,
            legend=False,
            ax=ax
        )
    
    # Tidy up legend and appearance
    ax.set_xlabel("Ave Distance to ASP311 or 304")
    ax.set_ylabel("Average FE Distance")
    ax.legend(bbox_to_anchor=(1.05, 1), loc="upper left", borderaxespad=0.)
    sns.despine()
    plt.tight_layout()
    plt.show()

from matplotlib.lines import Line2D

def orderplot(df,value,larger_better=True):
    if larger_better:
        df = df.loc[df.groupby(['cyp', 'ligand', 'charged'])[value].idxmax()].reset_index(drop=True)
    else:
        df = df.loc[df.groupby(['cyp', 'ligand', 'charged'])[value].idxmin()].reset_index(drop=True)

    # Make sure charged is boolean; if it's strings like "True"/"False", coerce:
    # df["charged"] = df["charged"].astype(bool)
    
    # Keep colors/shapes consistent across both panels
    cyp_order = sorted(df["cyp"].astype(str).unique())
    palette = dict(zip(cyp_order, sns.color_palette(n_colors=len(cyp_order))))
    markers = ["o", "s", "D", "^", "v", "P", "X", "*"]
    marker_map = {c: markers[i % len(markers)] for i, c in enumerate(cyp_order)}
        
    fig, axes = plt.subplots(1, 2, figsize=(12, 6), sharex=True,sharey=True, constrained_layout=True)
    
    panes = [
        ("charged = True", df[df["charged"] == True], axes[0]),
        ("charged = False", df[df["charged"] == False], axes[1]),
    ]
    
    for title, dsub, ax in panes:
        sns.scatterplot(
            data=dsub,
            x=value,
            y="ligand",
            hue="cyp",
            style="cyp",
            hue_order=cyp_order,
            style_order=cyp_order,
            markers=markers,        
            palette=palette,
            s=110,
            linewidth=0.8,
            edgecolor="black",
            ax=ax,
            legend=False,   # we'll add one combined legend later
        )
    
        # Label each point with last two chars of cyp
        for _, row in dsub.iterrows():
            ax.text(
                row[value],
                row["ligand"],
                str(row["cyp"])[-1:]+'\n',
                ha="center", va="bottom",
                fontsize=8,
                clip_on=True,
            )
    
        ax.set_title(title)
        ax.set_xlabel(value)
    
    # Shared y-label
    axes[0].set_ylabel("ligand")
    
    # === Manual legend ===
    legend_elements = [
        Line2D(
            [0], [0],
            marker=marker_map[c],
            color="w",
            label=c,
            markerfacecolor=palette[c],
            markeredgecolor="black",
            markersize=10,
            linewidth=0
        )
        for c in cyp_order
    ]
    
    fig.legend(
        handles=legend_elements,
        title="cyp",
        loc="upper center",
        ncol=3,
        bbox_to_anchor=(0.5, 1.12)
    )
    sns.despine()
    plt.show()

plot_scatter(df_min)

orderplot(df,'frac6fe',True) # fraction of time distance to FE is less than 6

orderplot(df,'avefe',False) # average distance to FE

orderplot(df,'avermsf',False) # average RMSF - how stable is the ligand pose

orderplot(df,'ASP_min',False) # minimum average distance to an aspartic (either 311 or 304) from charged atom

orderplot(prefdf,'frac6fe',True)

orderplot(prefdf,'avefe',False)

orderplot(prefdf,'avermsf',False)

orderplot(prefdf,'ASP_min',False)

charged_names = {
    'azelastine':'N52',
    'clemizole':'N44',
    'clomipramine':'N46',
    'clozapine_charged':'N43',
    'clozapine_uncharged':'N37',
    'eticlopride_charged':'N49',
    'eticlopride_uncharged':'N38',
    'fipexide_charged':'N49',
    'fipexide_uncharged':'N41',
    'prochlorperazine_charged':'N50',
    'prochlorperazine_uncharged':'N39',
    'sertraline_charged':'N38',
    'sertraline_uncharged':'N32',
    'terconazole_charged':'N68'
}

prefix='/net/galaxy/home/koes/spirhadi/cyp-project/worm/'
for label in df_min[df_min.charged].sort_values(by=['ligand','cyp']).label:
    vals = label.split('_')
    ligand = vals[1]
    if ligand in charged_names:
        analyze(prefix+label.replace('_','/'), charged_names[ligand],verbose=False)
    else:
        analyze(prefix+label.replace('_','/').replace('charged/1','charged_1').replace('charged/2','charged_2'), charged_names[f'{ligand}_charged'],verbose=False)

Average ligand RMSD: 3.10
Average ligand RMSF: 1.843
19 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 47.10% of simulation
The ligand atom with the shortest average distance to the HEME FE is C33.
Average distance of 4.75±0.48Å and a minimum distance of 3.43Å
Fraction within 6Å: 0.986
TYR104:OH ave dist 5.55±1.54Å to N52; min dist 3.14Å
ALA484:CB ave dist 6.23±1.50Å to N52; min dist 3.71Å
PHE213:CD1 ave dist 6.25±2.12Å to N52; min dist 3.30Å

Average ligand RMSD: 2.16
Average ligand RMSF: 1.597
5 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 97.93% of simulation
The ligand atom with the shortest average distance to the HEME FE is C34.
Average distance of 4.45±0.44Å and a minimum distance of 3.22Å
Fraction within 6Å: 1.000
ASP311:OD2 ave dist 3.43±0.86Å to N52; min dist 2.55Å
ASP311:OD1 ave dist 3.72±0.77Å to N52; min dist 2.57Å
ASP311:CG ave dist 3.75±0.64Å to N52; min dist 2.97Å

Average ligand RMSD: 2.20
Average ligand RMSF: 1.217
7 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 99.73% of simulation
The ligand atom with the shortest average distance to the HEME FE is C34.
Average distance of 4.98±0.34Å and a minimum distance of 3.82Å
Fraction within 6Å: 0.994
ASP311:OD1 ave dist 3.45±0.97Å to N52; min dist 2.57Å
ASP311:OD2 ave dist 3.57±0.92Å to N52; min dist 2.58Å
ASP311:CG ave dist 3.73±0.75Å to N52; min dist 3.05Å

Average ligand RMSD: 3.42
Average ligand RMSF: 2.756
5 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 99.83% of simulation
The ligand atom with the shortest average distance to the HEME FE is C28.
Average distance of 7.00±2.06Å and a minimum distance of 3.30Å
Fraction within 6Å: 0.333
ASP304:OD1 ave dist 2.75±0.11Å to N44; min dist 2.50Å
ASP304:CG ave dist 3.86±0.14Å to N44; min dist 3.31Å
ASP304:CA ave dist 4.25±0.23Å to N44; min dist 3.50Å

Average ligand RMSD: 3.40
Average ligand RMSF: 1.951
10 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 75.43% of simulation
The ligand atom with the shortest average distance to the HEME FE is C28.
Average distance of 4.48±0.71Å and a minimum distance of 3.34Å
Fraction within 6Å: 0.964
LEU307:O ave dist 5.37±0.89Å to N44; min dist 2.81Å
LEU307:C ave dist 5.71±0.61Å to N44; min dist 3.78Å
LEU307:CB ave dist 5.82±0.69Å to N44; min dist 3.54Å

Average ligand RMSD: 4.69
Average ligand RMSF: 1.620
14 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 88.93% of simulation
The ligand atom with the shortest average distance to the HEME FE is C28.
Average distance of 4.73±0.31Å and a minimum distance of 3.62Å
Fraction within 6Å: 1.000
LEU307:O ave dist 4.32±0.61Å to N44; min dist 2.61Å
ASP311:OD1 ave dist 4.65±2.31Å to N44; min dist 2.54Å
ASP311:CG ave dist 4.67±1.83Å to N44; min dist 3.05Å

Average ligand RMSD: 3.12
Average ligand RMSF: 1.476
19 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 95.03% of simulation
The ligand atom with the shortest average distance to the HEME FE is C28.
Average distance of 5.45±0.59Å and a minimum distance of 4.33Å
Fraction within 6Å: 0.881
ASP311:OD1 ave dist 3.71±1.04Å to N46; min dist 2.55Å
ASP311:OD2 ave dist 3.77±1.03Å to N46; min dist 2.55Å
ASP311:CG ave dist 3.83±0.72Å to N46; min dist 2.97Å

Average ligand RMSD: 2.20
Average ligand RMSF: 1.406
7 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 97.50% of simulation
The ligand atom with the shortest average distance to the HEME FE is C27.
Average distance of 4.61±0.38Å and a minimum distance of 3.59Å
Fraction within 6Å: 0.995
ASP311:OD1 ave dist 3.32±0.68Å to N46; min dist 2.55Å
ASP311:OD2 ave dist 3.37±0.72Å to N46; min dist 2.54Å
ASP311:CG ave dist 3.54±0.38Å to N46; min dist 2.95Å

Average ligand RMSD: 2.90
Average ligand RMSF: 1.446
8 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 99.00% of simulation
The ligand atom with the shortest average distance to the HEME FE is C27.
Average distance of 4.85±0.34Å and a minimum distance of 3.86Å
Fraction within 6Å: 0.996
ASP311:OD1 ave dist 3.63±0.86Å to N46; min dist 2.57Å
ASP311:OD2 ave dist 3.63±0.97Å to N46; min dist 2.52Å
ASP311:CG ave dist 3.73±0.48Å to N46; min dist 2.98Å

Average ligand RMSD: 3.89
Average ligand RMSF: 2.125
11 ligand clusters with 2.0Å radius (average linkage)
Top 3 clusters represent 88.37% of simulation
The ligand atom with the shortest average distance to the HEME FE is C26.
Average distance of 5.63±0.81Å and a minimum distance of 4.03Å
Fraction within 6Å: 0.668
ALA210:CB ave dist 5.86±1.72Å to N43; min dist 3.51Å
ASP311:OD1 ave dist 5.88±1.41Å to N43; min dist 2.68Å
ASP311:OD2 ave dist 5.89±1.32Å to N43; min dist 2.59Å

	label	cyp	ligand	model	charged	avermsf	nclusts	nclustsperc	featom	avefe	...	frac6fe	ASP311_OD1_ave	ASP311_OD1_std	ASP311_OD2_ave	ASP311_OD2_std	ASP304_OD1_ave	ASP304_OD1_std	ASP304_OD2_ave	ASP304_OD2_std	ASP_min
0	cyp35b1_clemizole_models_model1	cyp35b1	clemizole	models_model1	True	2.755680	5	99.833333	C28	7.004436	...	0.332667	12.616459	0.993289	12.221389	1.326094	2.748673	0.105565	4.737415	0.196022	2.748673
1	cyp35b1_clemizole_models_model2	cyp35b1	clemizole	models_model2	True	1.171633	2	100.000000	C28	8.624297	...	0.000000	12.323451	0.891017	12.432032	1.049660	2.734250	0.086308	4.750673	0.144463	2.734250
2	cyp35b1_clemizole_models_model3	cyp35b1	clemizole	models_model3	True	1.148177	5	99.833333	C31	10.216063	...	0.000000	11.960422	0.968025	12.329617	0.935384	2.773906	0.110201	4.846911	0.141818	2.773906
3	cyp35b1_clemizole_models_model4	cyp35b1	clemizole	models_model4	True	2.455778	31	49.366667	C25	7.946717	...	0.073667	5.784297	1.587585	5.414953	1.520463	10.922581	0.711395	11.892527	1.050971	5.414953
4	cyp35b1_clemizole_models_model5	cyp35b1	clemizole	models_model5	True	2.043223	10	82.966667	CL22	5.746361	...	0.709333	11.238187	0.960753	11.264184	1.014249	3.197480	0.820171	5.081813	0.874114	3.197480
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
251	cyp35b2_azelastine_uncharged_models_model2	cyp35b2	azelastine	uncharged_models_model2	False	2.751167	21	65.000000	C34	5.745116	...	0.714000	6.970463	2.497723	6.842302	2.511325	12.630244	1.157978	12.571135	1.374061	6.842302
252	cyp35b2_azelastine_uncharged_models_model3	cyp35b2	azelastine	uncharged_models_model3	False	2.522804	15	76.566667	O26	8.047019	...	0.000667	6.038128	1.467116	5.974527	1.672112	12.019222	0.905342	13.261692	1.292404	5.974527
253	cyp35b2_azelastine_uncharged_models_model4	cyp35b2	azelastine	uncharged_models_model4	False	2.294584	15	83.600000	C30	6.190050	...	0.663333	6.573103	1.606960	6.533184	1.553055	13.117672	0.994050	14.413677	0.892222	6.533184
254	cyp35b2_azelastine_uncharged_models_model5	cyp35b2	azelastine	uncharged_models_model5	False	2.445196	23	78.166667	CL27	5.142812	...	0.797667	8.006782	1.601858	7.929558	1.567985	11.750858	1.524333	13.338495	1.342090	7.929558
255	cyp35b2_azelastine_uncharged_models_model6	cyp35b2	azelastine	uncharged_models_model6	False	1.535395	7	94.766667	CL27	5.030561	...	0.958333	7.065642	1.322088	6.749130	1.271836	11.852220	0.676716	13.222108	0.610477	6.749130

	label	cyp	ligand	model	charged	avermsf	nclusts	nclustsperc	featom	avefe	...	frac6fe	ASP311_OD1_ave	ASP311_OD1_std	ASP311_OD2_ave	ASP311_OD2_std	ASP304_OD1_ave	ASP304_OD1_std	ASP304_OD2_ave	ASP304_OD2_std	ASP_min
0	cyp35b1_clemizole_models_model1	cyp35b1	clemizole	models_model1	True	2.755680	5	99.833333	C28	7.004436	...	0.332667	12.616459	0.993289	12.221389	1.326094	2.748673	0.105565	4.737415	0.196022	2.748673
1	cyp35b1_clemizole_models_model2	cyp35b1	clemizole	models_model2	True	1.171633	2	100.000000	C28	8.624297	...	0.000000	12.323451	0.891017	12.432032	1.049660	2.734250	0.086308	4.750673	0.144463	2.734250
2	cyp35b1_clemizole_models_model3	cyp35b1	clemizole	models_model3	True	1.148177	5	99.833333	C31	10.216063	...	0.000000	11.960422	0.968025	12.329617	0.935384	2.773906	0.110201	4.846911	0.141818	2.773906
3	cyp35b1_clemizole_models_model4	cyp35b1	clemizole	models_model4	True	2.455778	31	49.366667	C25	7.946717	...	0.073667	5.784297	1.587585	5.414953	1.520463	10.922581	0.711395	11.892527	1.050971	5.414953
9	cyp35b2_clemizole_models_model4	cyp35b2	clemizole	models_model4	True	1.951156	10	75.433333	C28	4.476287	...	0.964000	6.192553	2.396453	6.225269	2.412194	7.962295	2.226264	9.597702	2.197542	6.192553
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
247	cyp35b1_azelastine_uncharged_model4	cyp35b1	azelastine	uncharged_model4	False	1.468213	4	99.933333	C29	5.970015	...	0.606333	7.066720	1.400919	7.114590	1.385066	12.735864	0.612221	12.783843	0.602896	7.066720
249	cyp35b1_azelastine_uncharged_model6	cyp35b1	azelastine	uncharged_model6	False	1.540535	5	98.800000	C30	5.553536	...	0.847333	7.214012	1.666787	7.425996	1.644505	12.695609	0.519363	13.072401	0.722250	7.214012
250	cyp35b2_azelastine_uncharged_models_model1	cyp35b2	azelastine	uncharged_models_model1	False	1.515686	6	99.833333	C34	5.015954	...	0.991667	7.749970	1.063150	7.505898	0.959403	10.544401	1.629440	12.396356	1.573657	7.505898
251	cyp35b2_azelastine_uncharged_models_model2	cyp35b2	azelastine	uncharged_models_model2	False	2.751167	21	65.000000	C34	5.745116	...	0.714000	6.970463	2.497723	6.842302	2.511325	12.630244	1.157978	12.571135	1.374061	6.842302
253	cyp35b2_azelastine_uncharged_models_model4	cyp35b2	azelastine	uncharged_models_model4	False	2.294584	15	83.600000	C30	6.190050	...	0.663333	6.573103	1.606960	6.533184	1.553055	13.117672	0.994050	14.413677	0.892222	6.533184

Import packages and load data¶

Data¶

Plotting Code¶

Plots¶

Ignore poses that don't place a carbon closest to FE¶

Structures¶

cyp35b1_azelastine_models_model3

cyp35b2_azelastine_models_model1

cyp35b3_azelastine_models_model2

cyp35b1_clemizole_models_model1

cyp35b2_clemizole_models_model4

cyp35b3_clemizole_models_model6

cyp35b1_clomipramine_models_model2

cyp35b2_clomipramine_models_model6

cyp35b3_clomipramine_models_model6

cyp35b1_clozapine_charged_models_model3

cyp35b2_clozapine_charged_models_model3

cyp35b3_clozapine_charged_models_model2

cyp35b1_eticlopride_charged_models_model2

cyp35b2_eticlopride_charged_models_model2

cyp35b3_eticlopride_charged_models_model2

cyp35b1_fipexide_charged_models_model5

cyp35b2_fipexide_charged_models_model3

cyp35b3_fipexide_charged_models_model2

cyp35b1_prochlorperazine_charged_1_models_model4

cyp35b2_prochlorperazine_charged_2_models_model4

cyp35b3_prochlorperazine_charged_1_models_model3

cyp35b1_sertraline_charged_models_model6

cyp35b2_sertraline_charged_models_model5

cyp35b3_sertraline_charged_models_model5

cyp35b2_terconazole_charged_models_model2

cyp35b3_terconazole_charged_models_model1