diff --git a/workflow_peakcaller/scripts/multipoisson_peak_calling.py b/workflow_peakcaller/scripts/multipoisson_peak_calling.py
index d454e9a..de30169 100644
--- a/workflow_peakcaller/scripts/multipoisson_peak_calling.py
+++ b/workflow_peakcaller/scripts/multipoisson_peak_calling.py
@@ -43,6 +43,7 @@ def poisson_filter(edit_c_df):
 
     prefix = 'subregion_'
     region_sites['{}fraction'.format(prefix)] = (region_sites['{}conversions'.format(prefix)]-region_sites['edited_bases'])/(region_sites['{}coverage'.format(prefix)] - region_sites['target_bases'])
+    region_sites.replace([np.inf, -np.inf], np.nan, inplace=True)
     region_sites = region_sites.fillna(0)
     
     depth_windows = [0, 10, 20, 30, 40, 50, 1000000]
@@ -129,4 +130,4 @@ def output_blank():
 
 else:
     write('\nNo values, outputting empty file\n')
-    output_blank()
\ No newline at end of file
+    output_blank()
diff --git a/workflow_peakcaller/scripts/score_peaks.py b/workflow_peakcaller/scripts/score_peaks.py
index 5d78724..ef710e2 100644
--- a/workflow_peakcaller/scripts/score_peaks.py
+++ b/workflow_peakcaller/scripts/score_peaks.py
@@ -2,7 +2,7 @@
 import os
 import pandas as pd
 import sys    
-import pandas as pd
+import numpy as np
 from scipy.stats import nbinom
 
 parser = argparse.ArgumentParser(description='Add statistical score to peaks')
@@ -10,24 +10,21 @@
 parser.add_argument('peaks_with_scores', type=str)
 parser.add_argument('all_windows_with_fdr')
 
-
-
 args = parser.parse_args()
 peaks_with_edit_fraction = args.peaks_with_edit_fraction
 peaks_with_scores = args.peaks_with_scores
 all_windows_with_fdr = args.all_windows_with_fdr 
 
 all_windows = pd.read_csv(all_windows_with_fdr, sep='\t')
+all_windows.replace([np.inf, -np.inf], np.nan, inplace=True)
+all_windows = all_windows.fillna(0)
 background_rate = all_windows.subregion_fraction.mean()
 
-
 print("Input file specified: {}".format(peaks_with_edit_fraction))
 print("Input file specified: {}".format(peaks_with_scores))
-
 print("Background rate for negative binomial peak scoring calculated as: {}".format(background_rate))
 
 
-
 def get_confidence(conversions, coverage, frac=background_rate):
     # # Determine nbinom CDF
     # (nbinom.cdf(nbr_failures,nbr_successes,prob_success))