analyze_peak_window_sizes

Code

import os,sys
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import json

import subprocess

Code

path_to_project_dir = "/beagle3/haky/users/saideep/projects/Con_EnPACT/models"

modality_pcs = {
    "H3K27ac":{
        "Flu":1,
        "NI":1
    },
    "H3K27me3":{
        "Flu":2,
        "NI":1
    },
    "H3K4me1":{
        "Flu":4,
        "NI":2
    },
    "H3K4me3":{
        "Flu":2,
        "NI":2
    },
    "ATAC":{
        "Flu":3,
        "NI":3
    }
}

window_sizes = [2,4,8,16,32,64,128]

Code

corr_tables = []

for modality in modality_pcs.keys():
    for context in ["Flu","NI"]:
        for window_size in window_sizes:
            cur_proj_dir = os.path.join(path_to_project_dir,context+"_"+modality+"_ws"+str(window_size))
            cur_corr_table = pd.read_csv(os.path.join(cur_proj_dir,"intermediates","evaluate_training","corrs_df_for_subsets.tsv"), sep="\t")
            cur_corr_table["modality"] = modality
            cur_corr_table["context"] = context
            cur_corr_table["window_size"] = window_size

            corr_tables.append(cur_corr_table)

corr_table_all = pd.concat(corr_tables)
corr_table_all.head()

	data_subset	cor_id	cor_type	corrs	modality	context	window_size
0	train	EnPACT_pred_vs_ground_truth	spearman	0.763115	H3K27ac	Flu	2
1	train	log2_epi_ref_track_vs_ground_truth	spearman	0.437594	H3K27ac	Flu	2
2	train	EnPACT_pred_vs_ground_truth	pearson	0.788496	H3K27ac	Flu	2
3	train	log2_epi_ref_track_vs_ground_truth	pearson	0.439296	H3K27ac	Flu	2
4	valid	EnPACT_pred_vs_ground_truth	spearman	0.719843	H3K27ac	Flu	2

Code

import seaborn as sns
import matplotlib.ticker as ticker
sns.set_theme(style="whitegrid")

corr_table_enpact = corr_table_all[corr_table_all["cor_id"]=="EnPACT_pred_vs_ground_truth"]

min_corr = corr_table_enpact["corrs"].min() - 0.1
max_corr = corr_table_enpact["corrs"].max() + 0.1

for data_subset in ["valid"]:

    corr_table_enpact_subset = corr_table_enpact[corr_table_enpact["data_subset"]==data_subset]
    fg = sns.FacetGrid(data=corr_table_enpact_subset, col="context", row="modality", margin_titles=True, height=4, aspect=1)
    fg.map_dataframe(sns.lineplot, x="window_size", y="corrs", hue="cor_type", palette="deep", marker='o').add_legend().set(ylim=(min_corr,max_corr))
    for ax in fg.axes.flat:
        ax.yaxis.set_major_locator(ticker.MultipleLocator(0.05))


corr_table_enpact_smwindow = corr_table_enpact[corr_table_enpact["window_size"]<32]

for data_subset in ["valid"]:
    corr_table_enpact_smwindow_subset = corr_table_enpact_smwindow[corr_table_enpact_smwindow["data_subset"]==data_subset]
    fg = sns.FacetGrid(data=corr_table_enpact_smwindow_subset, col="context", row="modality", margin_titles=True, height=4, aspect=1)
    fg.map_dataframe(sns.lineplot, x="window_size", y="corrs", hue="cor_type", palette="deep", marker='o').add_legend().set(ylim=(min_corr,max_corr))
    for ax in fg.axes.flat:
        ax.yaxis.set_major_locator(ticker.MultipleLocator(0.05))

Code

corr_table_enpact_atac = corr_table_enpact[corr_table_enpact["modality"]=="H3K4me3"]
corr_table_enpact_atac[corr_table_enpact_atac["window_size"]==4]

	data_subset	cor_id	cor_type	corrs	modality	context	window_size
0	train	EnPACT_pred_vs_ground_truth	spearman	0.903147	H3K4me3	Flu	4
2	train	EnPACT_pred_vs_ground_truth	pearson	0.910854	H3K4me3	Flu	4
4	valid	EnPACT_pred_vs_ground_truth	spearman	0.878218	H3K4me3	Flu	4
6	valid	EnPACT_pred_vs_ground_truth	pearson	0.882030	H3K4me3	Flu	4
8	test	EnPACT_pred_vs_ground_truth	spearman	0.874550	H3K4me3	Flu	4
10	test	EnPACT_pred_vs_ground_truth	pearson	0.882604	H3K4me3	Flu	4
0	train	EnPACT_pred_vs_ground_truth	spearman	0.906249	H3K4me3	NI	4
2	train	EnPACT_pred_vs_ground_truth	pearson	0.913789	H3K4me3	NI	4
4	valid	EnPACT_pred_vs_ground_truth	spearman	0.881794	H3K4me3	NI	4
6	valid	EnPACT_pred_vs_ground_truth	pearson	0.887225	H3K4me3	NI	4
8	test	EnPACT_pred_vs_ground_truth	spearman	0.872527	H3K4me3	NI	4
10	test	EnPACT_pred_vs_ground_truth	pearson	0.882037	H3K4me3	NI	4