MACS3/IO/BedGraphIO.pyx

   1 # cython: language_level=3
   2 # cython: profile=True
   3 # Time-stamp: <2020-12-03 11:46:58 Tao Liu>
   4
   5 """Module Description:  IO Module for bedGraph file
   6
   7 This code is free software; you can redistribute it and/or modify it
   8 under the terms of the BSD License (see the file LICENSE included with
   9 the distribution).
  10 """
  11
  12 # ------------------------------------
  13 # python modules
  14 # ------------------------------------
  15 import io
  16
  17 from MACS3.Signal.BedGraph import bedGraphTrackI
  18
  19 # ------------------------------------
  20 # constants
  21 # ------------------------------------
  22
  23 # ------------------------------------
  24 # C lib
  25 # ------------------------------------
  26
  27 from libc.stdio cimport *
  28 from libc.stdlib cimport *
  29
  30 # ------------------------------------
  31 # Misc functions
  32 # ------------------------------------
  33
  34 # ------------------------------------
  35 # Classes
  36 # ------------------------------------
  37
  38 cdef class bedGraphIO:
  39     """File Parser Class for bedGraph File.
  40
  41     There are two assumptions in my bedGraphTrackI object:
  42
  43     1. Continuous: the next region should be after the previous one
  44     unless they are on different chromosomes;
  45
  46     2. Non-overlapping: the next region should never have overlaps
  47     with preceding region.
  48
  49     If any of the above two criteria is violated, parsering will fail.
  50     """
  51     cdef:
  52         str bedGraph_filename
  53
  54     def __init__ ( self, str bedGraph_filename ):
  55         """f must be a filename or a file handler.
  56
  57         """
  58         self.bedGraph_filename = bedGraph_filename
  59
  60     def build_bdgtrack (self, double baseline_value=0):
  61         """Use this function to return a bedGraphTrackI object.
  62
  63         baseline_value is the value to fill in the regions not defined
  64         in bedGraph. For example, if the bedGraph is like:
  65
  66         chr1  100 200  1
  67         chr1  250 350  2
  68
  69         Then the region chr1:200..250 should be filled with
  70         baseline_value. Default of baseline_value is 0.
  71         """
  72         cdef bytes i
  73
  74         data = bedGraphTrackI(baseline_value=baseline_value)
  75         add_func = data.add_loc
  76         # python open file
  77         bedGraph_file = open( self.bedGraph_filename, "rb" )
  78
  79         for i in bedGraph_file:
  80             if i.startswith(b"track"):
  81                 continue
  82             elif i.startswith(b"#"):
  83                 continue
  84             elif i.startswith(b"browse"):
  85                 continue
  86             else:
  87                 fs = i.split()
  88                 add_func(fs[0],atoi(fs[1]),atoi(fs[2]),atof(fs[3]))
  89
  90         bedGraph_file.close()
  91         return data
  92