trim random primer and store sequence in rS tag #54.

singlecellmultiomics/modularDemultiplexer/baseDemultiplexMethods.py

@@ -325,6 +325,7 @@ def __init__(self,
         barcodeRead=0, barcodeStart = 6, barcodeLength=8,
          barcodeFileParser=None, barcodeFileAlias=None, indexFileParser=None,
          indexFileAlias = 'illumina_merged_ThruPlex48S_RP',
+         random_primer_read = None, random_primer_length = 6,
          **kwargs ):
         self.description=''
         self.barcodeFileAlias = barcodeFileAlias
@@ -340,7 +341,10 @@ def __init__(self,
         self.barcodeLength = barcodeLength
         self.autoDetectable = False
 
-        self.sequenceCapture = [slice(None) , slice(None) ] # ranges
+        self.random_primer_read = random_primer_read
+        self.random_primer_length = random_primer_length
+
+        self.sequenceCapture = [slice(None) , slice(None) ] # ranges to capture for read 1 and read 2
         if umiLength==0:
             # Barcode only
             if barcodeStart!=0:
@@ -353,6 +357,16 @@ def __init__(self,
                 raise NotImplementedError('Complicated slice where we need to capture around a region')
             self.sequenceCapture[barcodeRead] =     slice( barcodeLength+umiLength, None)
 
+        if random_primer_read!=None:
+            if self.sequenceCapture[random_primer_read].stop is not None:
+                raise NotImplementedError()
+            self.sequenceCapture[random_primer_read] = slice(
+                    self.sequenceCapture[random_primer_read].start,
+                    -random_primer_length,
+                    self.sequenceCapture[random_primer_read].step
+            )
+            self.random_primer_slice = slice(-random_primer_length,None,None)
+
     def __repr__(self):
         return f'{self.longName} bc: {self.barcodeStart}:{self.barcodeLength}, umi: {self.umiStart}:{self.umiLength} {self.description}'
 
@@ -376,6 +390,9 @@ def demultiplex(self, records, **kwargs):
         if barcodeIdentifier is None:
             raise NonMultiplexable(f'bc:{rawBarcode}_not_matching_{self.barcodeFileAlias}')
 
+        random_primer = None
+        if self.random_primer_read!=None:
+            random_primer = records[self.random_primer_read].sequence[self.random_primer_slice]
         if self.umiLength!=0:
             umi = records[self.umiRead].sequence[self.umiStart:self.umiStart+self.umiLength]
             umiQual = records[self.umiRead].qual[self.umiStart:self.umiStart+self.umiLength]
@@ -406,7 +423,11 @@ def demultiplex(self, records, **kwargs):
                         'BC':barcode
             })
             #tr.addTagByTag('hd', hammingDistance, isPhred=False)
-
+            if random_primer is not None:
+                tr.addTagByTag('rP',
+                    random_primer,
+                    isPhred=False,
+                    make_safe=False)
 
             tr.addTagByTag('QT', barcodeQual, isPhred=True)
 

singlecellmultiomics/modularDemultiplexer/demultiplexModules/CELSeq1.py

@@ -8,8 +8,9 @@ def __init__(self, barcodeFileParser,  **kwargs  ):
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 8, umiLength=4,
 		barcodeRead=0, barcodeStart = 0, barcodeLength=8,
+		random_primer_read=1, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser, **kwargs   )
 		self.shortName = 'CS1C8U4'
 		self.longName = 'CELSeq 1, CB: 8bp, UMI: 4bp'
 		self.autoDetectable = True
-		self.description = 'R1 starts with a 8bp cell barcode followed by a 4bp UMI'
+		self.description = 'R1 starts with a 8bp cell barcode followed by a 4bp UMI. R2 ends with a 6bp random primer'

singlecellmultiomics/modularDemultiplexer/demultiplexModules/CELSeq2.py

@@ -7,11 +7,12 @@ def __init__(self, barcodeFileParser,  **kwargs  ):
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 0, umiLength=6,
 		barcodeRead=0, barcodeStart = 6, barcodeLength=8,
+		random_primer_read=1, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs   )
 		self.shortName = 'CS2C8U6'
 		self.longName = 'CELSeq 2, CB: 8bp, UMI: 6bp'
 		self.autoDetectable = True
-		self.description = 'R1 starts with a 6bp UMI  followed by a 8bp cell barcode'
+		self.description = 'R1 starts with a 6bp UMI  followed by a 8bp cell barcode. R2 ends with a 6bp random primer'
 
 # Reversed case:
 class CELSeq2_c8_u6_swapped_reads(UmiBarcodeDemuxMethod):
@@ -20,11 +21,12 @@ def __init__(self, barcodeFileParser,  **kwargs  ):
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=1, umiStart = 0, umiLength=6,
 		barcodeRead=1, barcodeStart = 6, barcodeLength=8,
+		random_primer_read=0, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs   )
 		self.shortName = 'CS2C8U6S'
-		self.longName = 'CELSeq 2, CB: 8bp, UMI: 6bp'
+		self.longName = 'CELSeq 2, CB: 8bp, UMI: 6bp, RP: 6bp'
 		self.autoDetectable = True
-		self.description = 'R2 starts with a 6bp UMI  followed by a 8bp cell barcode'
+		self.description = 'R2 starts with a 6bp UMI  followed by a 8bp cell barcode. R1 ends with a 6bp random primer'
 
 
 # Cell seq 2 with 8bp UMI
@@ -33,12 +35,13 @@ def __init__(self, barcodeFileParser,  **kwargs  ):
 		self.barcodeFileAlias = 'celseq2'
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 0, umiLength=8,
+		random_primer_read=1, random_primer_length=6,
 		barcodeRead=0, barcodeStart = 8, barcodeLength=8,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs   )
 		self.shortName = 'CS2C8U8'
 		self.longName = 'CELSeq 2, CB: 8bp, UMI: 8bp'
 		self.autoDetectable = True
-		self.description = 'R1 starts with a longer 8bp UMI  followed by a 8bp cell barcode'
+		self.description = 'R1 starts with a longer 8bp UMI  followed by a 8bp cell barcode. R2 ends with a 6bp random primer'
 
 # Reversed case:
 class CELSeq2_c8_u8(UmiBarcodeDemuxMethod):
@@ -47,18 +50,20 @@ def __init__(self, barcodeFileParser,  **kwargs  ):
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=1, umiStart = 0, umiLength=8,
 		barcodeRead=1, barcodeStart = 8, barcodeLength=8,
+		random_primer_read=0, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs   )
 		self.shortName = 'CS2C8U8S'
 		self.longName = 'CELSeq 2, CB: 8bp, UMI: 8bp'
 		self.autoDetectable = True
-		self.description = 'R2 starts with a longer 8bp UMI  followed by a 8bp cell barcode'
+		self.description = 'R2 starts with a longer 8bp UMI  followed by a 8bp cell barcode. R1 ends with a 6bp primer'
 
 class CELSeq2_c8_u8_NNLAIII(UmiBarcodeDemuxMethod):
 	def __init__(self, barcodeFileParser,  **kwargs  ):
 		self.barcodeFileAlias = 'celseq2_noNla'
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 0, umiLength=8,
 		barcodeRead=0, barcodeStart = 8, barcodeLength=8,
+		random_primer_read=1, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs   )
 		self.shortName = 'CS2C8U8NNLA'
 		self.longName = 'CELSeq 2, CB: 8bp, UMI: 8bp, NLAIII free'

singlecellmultiomics/modularDemultiplexer/demultiplexModules/NLAIII.py

@@ -6,12 +6,14 @@ def __init__(self, barcodeFileParser,  **kwargs ):
 		self.barcodeFileAlias = 'lennart96NLA'
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 0, umiLength=3,
+		random_primer_read=1, random_primer_length=6,
 		barcodeRead=0, barcodeStart = 3, barcodeLength=8,
+
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs  )
 		self.shortName = 'NLAIII96C8U3'
-		self.longName = 'NLAIII, 96well CB: 8bp UMI: 3bp'
+		self.longName = 'NLAIII, 96well CB: 8bp UMI: 3bp RP: 6bp'
 		self.autoDetectable = True
-		self.description = '96 well format. 3bp umi followed by 8bp barcode'
+		self.description = '96 well format. 3bp umi followed by 8bp barcode. R2 ends with a 6bp random primer'
 
 
 
@@ -22,11 +24,12 @@ def __init__(self, barcodeFileParser,  **kwargs ):
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 0, umiLength=3,
 		barcodeRead=0, barcodeStart = 3, barcodeLength=8,
+		random_primer_read=1, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs  )
 		self.shortName = 'NLAIII384C8U3'
-		self.longName = 'NLAIII, 384well CB: 8bp UMI: 3bp'
+		self.longName = 'NLAIII, 384well CB: 8bp UMI: 3bp RP:6bp'
 		self.autoDetectable = True
-		self.description = '384 well format. 3bp umi followed by 8bp barcode'
+		self.description = '384 well format. 3bp umi followed by 8bp barcode. R2 ends with a 6bp random primer'
 
 	def demultiplex(self, records, **kwargs):
 		if kwargs.get('probe') and records[0].sequence[self.barcodeLength+ self.umiLength : self.barcodeLength+ self.umiLength+4]!='CATG':

singlecellmultiomics/modularDemultiplexer/demultiplexModules/scCHIC.py

@@ -8,11 +8,12 @@ def __init__(self, barcodeFileParser,  **kwargs ):
 		UmiBarcodeDemuxMethod.__init__(self,
 		umiRead=0, umiStart = 0, umiLength=3,
 		barcodeRead=0, barcodeStart = 3, barcodeLength=8,
+		random_primer_read=1, random_primer_length=6,
 	 	barcodeFileAlias = self.barcodeFileAlias ,barcodeFileParser=barcodeFileParser,  **kwargs  )
 		self.shortName = 'scCHIC384C8U3'
-		self.longName = 'Single cell CHIC, 384well CB: 8bp UMI: 3bp'
+		self.longName = 'Single cell CHIC, 384well CB: 8bp UMI: 3bp, RP: 6BP'
 		self.autoDetectable = True
-		self.description = '384 well format. 3bp umi followed by 8bp barcode and a single A'
+		self.description = '384 well format. 3bp umi followed by 8bp barcode and a single A. R2 ends with a 6bp random primer'
 
 		self.sequenceCapture[0] = slice( self.barcodeLength+ self.umiLength + 1, None) # dont capture the first base