pileup support for FRAG format done

Author: taoliu

MACS3/Commands/pileup_cmd.py

@@ -4,7 +4,7 @@
 under the terms of the BSD License (see the file LICENSE included with
 the distribution).
 """
-# Time-stamp: <2024-10-02 16:51:15 Tao Liu>
+# Time-stamp: <2024-11-30 00:04:48 Tao Liu>
 # ------------------------------------
 # python modules
 # ------------------------------------
@@ -16,7 +16,8 @@
 # ------------------------------------
 # from MACS3.Utilities.Constants import *
 from MACS3.Utilities.OptValidator import opt_validate_pileup
-from MACS3.Signal.Pileup import pileup_and_write_se, pileup_and_write_pe
+from MACS3.Signal.Pileup import pileup_and_write_se
+from MACS3.IO.BedGraphIO import bedGraphIO
 # ------------------------------------
 # Main function
 # ------------------------------------
@@ -35,7 +36,7 @@ def run(o_options):
     # error = options.error
 
     # 0 output arguments
-    options.PE_MODE = options.format in ('BAMPE', 'BEDPE')
+    options.PE_MODE = options.format in ('BAMPE', 'BEDPE', 'FRAG')
 
     # 0 prepare output file
     outfile = os.path.join(options.outdir, options.outputfile).encode()
@@ -51,8 +52,9 @@ def run(o_options):
         t0 = treat.total                          # total fragments
         info("# total fragments/pairs in alignment file: %d" % (t0))
         info("# Pileup paired-end alignment file.")
-        pileup_and_write_pe(treat, outfile)
-
+        bdg = treat.pileup_bdg()
+        bdgio = bedGraphIO(outfile, data=bdg)
+        bdgio.write_bedGraph(trackline=False)
     else:
         (tsize, treat) = load_tag_files_options(options)
 

MACS3/IO/BedGraphIO.py

@@ -1,6 +1,6 @@
 # cython: language_level=3
 # cython: profile=True
-# Time-stamp: <2024-10-08 10:07:47 Tao Liu>
+# Time-stamp: <2024-11-30 00:04:35 Tao Liu>
 
 """Module Description:  IO Module for bedGraph file
 
@@ -56,10 +56,10 @@ class bedGraphIO:
 
     If any of the above two criteria is violated, parsering will fail.
     """
-    bedGraph_filename = cython.declare(str, visibility='public')
+    bedGraph_filename = cython.declare(bytes, visibility='public')
     data = cython.declare(object, visibility='public')
 
-    def __init__(self, bedGraph_filename: str, data=None):
+    def __init__(self, bedGraph_filename: bytes, data=None):
         """f must be a filename or a file handler.
 
         """

MACS3/IO/Parser.py

@@ -1,7 +1,7 @@
 # cython: language_level=3
 # cython: profile=True
 # cython: linetrace=True
-# Time-stamp: <2024-10-22 10:25:23 Tao Liu>
+# Time-stamp: <2024-11-29 23:30:03 Tao Liu>
 
 """Module for all MACS Parser classes for input. Please note that the
 parsers are for reading the alignment files ONLY.
@@ -1528,6 +1528,10 @@ def pe_parse_line(self, thisline: bytes):
             raise Exception("Less than 5 columns found at this line: %s\n" %
                             thisline)
 
+    @cython.ccall
+    def build_petrack(self):
+        return self.build_petrack2()
+
     @cython.ccall
     def build_petrack2(self):
         """Build PETrackII from all lines.

MACS3/Signal/PairedEndTrack.py

@@ -1,6 +1,6 @@
 # cython: language_level=3
 # cython: profile=True
-# Time-stamp: <2024-10-15 15:56:00 Tao Liu>
+# Time-stamp: <2024-11-29 23:37:25 Tao Liu>
 
 """Module for filter duplicate tags from paired-end data
 
@@ -749,6 +749,8 @@ def add_loc(self,
                 self.buf_size[chromosome] += self.buffer_size
                 self.locations[chromosome].resize((self.buf_size[chromosome]),
                                                   refcheck=False)
+                self.barcodes[chromosome].resize((self.buf_size[chromosome]),
+                                                 refcheck=False)                
             self.locations[chromosome][i] = (start, end, count)
             self.barcodes[chromosome][i] = bn
             self.size[chromosome] = i + 1

MACS3/Utilities/OptValidator.py

@@ -1,4 +1,4 @@
-# Time-stamp: <2024-11-29 22:11:31 Tao Liu>
+# Time-stamp: <2024-11-29 23:48:10 Tao Liu>
 """Module Description
 
 This code is free software; you can redistribute it and/or modify it
@@ -104,6 +104,7 @@ def opt_validate_callpeak(options):
     # setting it as 'all'
     if options.format == 'FRAG' and options.keepduplicates != "all":
         logger.warning("Since the format is 'FRAG', `--keep-dup` will be set as 'all'.")
+        options.keepduplicates = "all"
 
     if options.extsize < 1:
         logger.error("--extsize must >= 1!")
@@ -537,7 +538,7 @@ def opt_validate_pileup(options):
     elif options.format == "BEDPE":
         options.parser = BEDPEParser
     elif options.format == "FRAG":
-        options.parser = FragParser        
+        options.parser = FragParser
     else:
         logger.error("Format \"%s\" cannot be recognized!" % (options.format))
         sys.exit(1)