From b3c92fbb6a0511d507e47a472a2c5c23c89ae69d Mon Sep 17 00:00:00 2001
From: jsidhom1 <jsidhom1@jhmi.edu>
Date: Tue, 8 Dec 2020 15:55:09 -0500
Subject: [PATCH] feature_updates

---
 .../supervised/HIV_analysis/reference.py      | 44 +++++++++++++++++++
 1 file changed, 44 insertions(+)
 create mode 100644 ancillary_analysis/supervised/HIV_analysis/reference.py

diff --git a/ancillary_analysis/supervised/HIV_analysis/reference.py b/ancillary_analysis/supervised/HIV_analysis/reference.py
new file mode 100644
index 0000000..7b68151
--- /dev/null
+++ b/ancillary_analysis/supervised/HIV_analysis/reference.py
@@ -0,0 +1,44 @@
+import pickle
+import pandas as pd
+import numpy as np
+import seaborn as sns
+import matplotlib.pyplot as plt
+from scipy.stats import fisher_exact
+
+with open('screen.pkl','rb') as f:
+    [df_auc,sequences,predicted] = pickle.load(f)
+
+pw = pd.read_csv('hiv_ref_pw.csv')
+preds = []
+for row in pw.iterrows():
+    tcr = row[1]['TCR']
+    pep = row[1]['Peptide']
+    u = np.where(df_auc['epitope']==pep)[0][0]
+    preds.append(np.mean(predicted[u][np.where(np.isin(sequences[u],tcr))[0]]))
+
+pw['preds'] = preds
+pw['preds'] = pw['preds'].round(3)
+pw.to_csv('val_preds.csv',index=False)
+
+all_pred = np.hstack(predicted)
+df_plot = pd.DataFrame()
+df_plot['preds'] = np.hstack([preds,all_pred])
+df_plot['label'] = np.hstack([['validated+']*len(preds),['background']*len(all_pred)])
+sns.violinplot(data=df_plot[df_plot['label']=='background'],x='label',y='preds',cut=0)
+sns.swarmplot(data=df_plot[df_plot['label']=='validated+'],x='label',y='preds',color='red',size=8,alpha=0.75)
+plt.xlabel('')
+plt.ylabel('Prediction Value',fontsize=16)
+plt.xticks([])
+plt.yticks()
+plt.show()
+plt.savefig('val_dist.png')
+
+thresh = 0.95
+x = np.array([[np.sum(all_pred < thresh)-1,np.sum(all_pred >= thresh)-17],
+     [1,17]])
+
+_,p_val = fisher_exact(x)
+enrichment = (x[1,1]/np.sum(x[:,1]))/(np.sum(x[1,:])/np.sum(x))
+
+
+