Fix asynchrony in input file splitting CRITICAL FIX

python_only_tool/DemultiplexUsingBarcodes_New_V2.py

@@ -201,7 +201,7 @@ def return_fastq(self):
     ######
     bin_counter = 0
     for i in range(0,int(linesInInputFastq),int(binIterator)):
-        # Create dictinaries used to store the parsed read information from the fastq files
+        # Create dictionaries used to store the parsed read information from the fastq files
         readsF = {}
         readsR = {}
         readsF_BC_UMI_dict = {}
@@ -241,13 +241,14 @@ def return_fastq(self):
                     lineQuality=str(line[0:].rstrip())
                     completeReadCounter += 1
                 if (completeReadCounter == 3): 
+                    read_counter += 1
                     processedRead = FastQRead(name = lineName, \
                         read = lineRead, \
                         quality = lineQuality, \
                         lineNumber = read_counter)
+                    print("Adding Fwd Read Key " + str(str(bin_counter) + "_" + str(read_counter)))
                     readsF[str(str(bin_counter) + "_" + str(read_counter))]=processedRead
                     completeReadCounter = 0
-                    read_counter += 1
                 if (starterF == 1):
                     line_ct1 += 1
 
@@ -293,16 +294,17 @@ def return_fastq(self):
                     lineQuality=str(line[0:].rstrip())
                     completeReadCounter += 1
                 if (completeReadCounter == 3):     
+                    read_counter += 1 # The read counter should progress even if a read does not satisfy the following criteria for being retained.
                     if len(filteredBarcode1) == 0:  # The following if statments break the loop if a barcode does not pass the HD <=1 filter
-                        line_ct1 += 1  # If the barcode observed does not match a barcode in our greenlist we escape the loop, count the line and reset the complete read counter.
+                        #line_ct1 += 1  # If the barcode observed does not match a barcode in our greenlist we escape the loop, count the line and reset the complete read counter.
                         completeReadCounter = 0
                         continue
                     elif len(filteredBarcode2) == 0:
-                        line_ct1 += 1
+                        #line_ct1 += 1
                         completeReadCounter = 0
                         continue
                     elif len(filteredBarcode3) == 0:
-                        line_ct1 += 1
+                        #line_ct1 += 1
                         completeReadCounter = 0
                         continue
                     else:
@@ -314,9 +316,9 @@ def return_fastq(self):
                         barcode2 = filteredBarcode2[0], \
                         barcode3 = filteredBarcode3[0], \
                         umi = lineReadUMI)
+                        print("Adding Rev Read Key " + str(str(bin_counter) + "_" + str(read_counter)))
                         readsR[str(str(bin_counter) + "_" + str(read_counter))]=processedRead
                         completeReadCounter = 0  # Reset the complete read counter to 0 because a read has been completely extracted and stored in the dictionary.      
-                    read_counter += 1 # The read counter should progress even if a read does not satisfy the criteria for being retained. 
                 if (starterR == 1):
                     line_ct1 += 1  # There are 4 lines for each read in the fastq file. The line counter needs to progress even if no "if" statements are satisfied.
 
@@ -329,6 +331,7 @@ def return_fastq(self):
         # while the forward read contains the gene expression information.
         # So we need to create a read that contains the barcode information appended to the read name 
         # and the forward read information stored as the actual read.
+        print("beginning transfer of barcodes")
         for key in readsR.keys():
             readF_BC_UMI = barcodeRead(name = readsF[key].name, \
                 read = readsF[key].read, \