Merge pull request #13 from a-slide/dev

Dev

NanopolishComp/Eventalign_collapse.py

@@ -13,15 +13,18 @@
 # Standard library imports
 import multiprocessing as mp
 from time import time
-from collections import OrderedDict
+from collections import *
 import traceback
+import datetime
 
 # Third party imports
 import numpy as np
 from tqdm import tqdm
 
 # Local imports
-from NanopolishComp.common import file_readable, dir_writable, NanopolishCompError
+from NanopolishComp.common import *
+from NanopolishComp import __version__ as package_version
+from NanopolishComp import __name__ as package_name
 
 #~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~LOGGING INFO~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#
 import logging
@@ -47,7 +50,7 @@ def __init__ (self,
         * input_fn
             Path to a nanopolish eventalign tsv output file.
         * outdir
-            Path to the output folder
+            Path to the output folder (will be created if it does exist yet)
         * outprefix
             text outprefix for all the files generated
         * max_reads
@@ -65,21 +68,39 @@ def __init__ (self,
             Reduce verbosity
         """
 
+        # Save init options in dict for later
+        kwargs = locals()
+
         # Define overall verbose level
         self.log = logging.getLogger()
         if verbose:
             self.log.setLevel (logging.DEBUG)
         elif quiet:
             self.log.setLevel (logging.WARNING)
+        else:
+            self.log.setLevel (logging.INFO)
+
+        # Collect args in dict for log report
+        self.option_d = OrderedDict()
+        self.option_d["package_name"] = package_name
+        self.option_d["package_version"] = package_version
+        self.option_d["timestamp"] = str(datetime.datetime.now())
+        for i, j in kwargs.items():
+            if i != "self":
+                self.option_d[i]=j
+        self.log.debug ("Options summary")
+        self.log.debug (dict_to_str(self.option_d))
 
         # Verify parameters validity
         self.log.info ("Checking arguments")
+        # Try to read input file if not a stream
         self.log.debug("\tTesting input file readability")
         if input_fn != 0 and not file_readable (input_fn):
             raise IOError ("Cannot read input file")
-        self.log.info("Testing output dir writability")
-        if not dir_writable (outdir):
-            raise IOError ("Cannot write output file in indicated folder. Create the output folder if it does not exist yet")
+        # Try to create output folder
+        self.log.debug("\tCreating output folder")
+        mkdir(outdir, exist_ok=True)
+        # Check other args
         self.log.debug("\tChecking number of threads")
         if threads < 3:
             raise ValueError ("At least 3 threads required")
@@ -128,6 +149,11 @@ def __init__ (self,
             self.log.warning ("\nAn error occured. All processes were killed\n")
             raise E
 
+    def __repr__ (self):
+        m = "General options:\n"
+        m+=dict_to_str(self.option_d)
+        return m
+
     #~~~~~~~~~~~~~~~~~~~~~~~~~~~~~PRIVATE METHODS~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#
     def _split_reads (self, in_q, error_q):
         """
@@ -307,6 +333,11 @@ def _write_output (self, out_q, error_q):
                 # Flag last line
                 data_fp.write ("#\n")
 
+            # Open log file
+            log_fn = os.path.join(self.outdir, self.outprefix+"_eventalign_collapse.log")
+            with open (log_fn, "w") as log_fp:
+                log_fp.write (str(self))
+
         # Manage exceptions and deal poison pills
         except Exception:
             error_q.put (NanopolishCompError(traceback.format_exc()))

NanopolishComp/Freq_meth_calculate.py

@@ -6,12 +6,15 @@
 # Standard library imports
 from collections import *
 import csv
+import datetime
 
 # Third party imports
 from tqdm import tqdm
 
 # Local imports
 from NanopolishComp.common import *
+from NanopolishComp import __version__ as package_version
+from NanopolishComp import __name__ as package_name
 
 #~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~LOGGING INFO~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#
 import logging
@@ -37,7 +40,7 @@ def __init__ (self,
         * input_fn
             Path to a nanopolish call_methylation tsv output file
         * outdir
-            Path to the output folder
+            Path to the output folder (will be created if it does exist yet)
         * outprefix
             text outprefix for all the files generated
         * min_llr
@@ -56,21 +59,38 @@ def __init__ (self,
             Reduce verbosity
         """
 
+        # Save init options in dict for later
+        kwargs = locals()
+
         # Define overall verbose level
         self.log = logging.getLogger()
         if verbose:
             self.log.setLevel (logging.DEBUG)
         elif quiet:
             self.log.setLevel (logging.WARNING)
+        else:
+            self.log.setLevel (logging.INFO)
+
+        # Collect args in dict for log report
+        self.option_d = OrderedDict()
+        self.option_d["package_name"] = package_name
+        self.option_d["package_version"] = package_version
+        self.option_d["timestamp"] = str(datetime.datetime.now())
+        for i, j in kwargs.items():
+            if i != "self":
+                self.option_d[i]=j
+        self.log.debug ("Options summary")
+        self.log.debug (dict_to_str(self.option_d))
 
         # Verify parameters validity
         self.log.warning ("## Checking arguments ##")
-        self.log.info("Test input file readability")
+        # Try to read input file if not a stream
+        self.log.debug("\tTesting input file readability")
         if input_fn != 0 and not file_readable (input_fn):
             raise IOError ("Cannot read input file")
-        self.log.info("Testing output dir writability")
-        if not dir_writable (outdir):
-            raise IOError ("Cannot write output file in indicated folder. Create the output folder if it does not exist yet")
+        # Try to create output folder
+        self.log.debug("\tCreating output folder")
+        mkdir(outdir, exist_ok=True)
 
         if motif == "cpg":
             motif_seq = "CG"
@@ -144,7 +164,7 @@ def _parse_methylation_calls(self, fn, min_llr=2.5, min_depth=10, min_meth_freq=
 
         # Print read level counter summary
         self.log.debug ("Read sites summary")
-        self.log.debug (counter_to_str(self.site_c))
+        self.log.debug (dict_to_str(self.site_c))
 
         self.log.info ("Filtering out positions with low coverage or methylation frequency")
         filtered_sites_d = OrderedDict()
@@ -164,16 +184,17 @@ def _parse_methylation_calls(self, fn, min_llr=2.5, min_depth=10, min_meth_freq=
 
         # Print genomic positions level counter summary
         self.log.debug ("Genomic positions summary")
-        self.log.debug (counter_to_str(self.pos_c))
+        self.log.debug (dict_to_str(self.pos_c))
 
         return filtered_sites_d
 
     def __repr__ (self):
-        m = ""
+        m = "General options:\n"
+        m+=dict_to_str(self.option_d)
         m+="Read sites summary:\n"
-        m+=counter_to_str(self.site_c)
+        m+=dict_to_str(self.site_c)
         m+="Genomic positions summary:\n"
-        m+=counter_to_str(self.pos_c)
+        m+=dict_to_str(self.pos_c)
         return m
 
     def _write_output (self, sites_d, outdir, outprefix):
@@ -202,7 +223,6 @@ def _write_output (self, sites_d, outdir, outprefix):
         self.log.info("Writing log file")
         fn = os.path.join(outdir, outprefix+"_freq_meth_calculate.log")
         with open (fn, "w") as fp:
-            # Write header
             fp.write (str(self))
 
 #~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~HELPER CLASS~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~#

NanopolishComp/__init__.py

@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 
 # Define self package variable
-__version__ = "0.6.1"
+__version__ = "0.6.2"
 __description__ = 'NanopolishComp is a Python3 package for downstream analyses of Nanopolish output files'
 
 # Collect info in a dictionary for setup.py

NanopolishComp/__main__.py

@@ -30,7 +30,7 @@ def main(args=None):
     subparser_ec.set_defaults(func=Eventalign_collapse_main)
     subparser_ec_io = subparser_ec.add_argument_group("Input/Output options")
     subparser_ec_io.add_argument("-i", "--input_fn", default=0, help="Path to a nanopolish eventalign tsv output file. If '0' read from std input (default: %(default)s)")
-    subparser_ec_io.add_argument("-o", "--outdir", type=str, default="./", help="Path to the output folder (default: %(default)s)")
+    subparser_ec_io.add_argument("-o", "--outdir", type=str, default="./", help="Path to the output folder (will be created if it does exist yet) (default: %(default)s)")
     subparser_ec_io.add_argument("-p", "--outprefix", type=str, default="out", help="text outprefix for all the files generated (default: %(default)s)")
     subparser_ec_rp = subparser_ec.add_argument_group("Run parameters options")
     subparser_ec_rp.add_argument("-s", "--write_samples", default=False, action='store_true', help="If given, will write the raw sample if nanopolish eventalign was ran with --samples option (default: %(default)s)")
@@ -44,7 +44,7 @@ def main(args=None):
     subparser_fm.set_defaults(func=Freq_meth_calculate_main)
     subparser_fm_io = subparser_fm.add_argument_group("Input/Output options")
     subparser_fm_io.add_argument("-i", "--input_fn", default=0, help="Path to a nanopolish call_methylation tsv output file. If not specified read from std input")
-    subparser_fm_io.add_argument("-o", "--outdir", type=str, default="./", help="Path to the output folder (default: %(default)s)")
+    subparser_fm_io.add_argument("-o", "--outdir", type=str, default="./", help="Path to the output folder (will be created if it does exist yet) (default: %(default)s)")
     subparser_fm_io.add_argument("-p", "--outprefix", type=str, default="out", help="text outprefix for all the files generated (default: %(default)s)")
     subparser_fm_fo = subparser_fm.add_argument_group("Filtering options")
     subparser_fm_fo.add_argument("-l", "--min_llr", type=float, default=2.5, help="Log likelihood ratio threshold (default: %(default)s)")

NanopolishComp/common.py

@@ -23,6 +23,13 @@ def dir_writable (fn, **kwargs):
         fn = os.path.dirname(fn)
     return os.path.dirname(fn) and os.access (fn, os.W_OK)
 
+def mkdir (fn, exist_ok=False):
+    """ Create directory recursivelly. Raise IO error if path exist or if error at creation """
+    try:
+        os.makedirs (fn, exist_ok=exist_ok)
+    except:
+        raise NanopolishCompError ("Error creating output folder `{}`".format(fn))
+
 def numeric_cast_dict (d):
     """Cast str values to integer or float from a dict """
     for k, v in d.items():
@@ -48,11 +55,15 @@ def find_subseq_index (seq, subseq):
         yield i
         i = seq.find(subseq, i+1)
 
-def counter_to_str (c):
-    """ Transform a counter dict to a tabulated str """
+def dict_to_str (c):
+    """ Transform a dict to a tabulated str """
     m = ""
-    for i, j in c.most_common():
-        m += "\t{}: {:,}\n".format(i, j)
+    if type(c) == Counter:
+        for i, j in c.most_common():
+            m += "\t{}: {:,}\n".format(i, j)
+    else:
+        for i, j in c.items():
+            m += "\t{}: {}\n".format(i, j)
     return m
 
 def jhelp (f:"python function or method"):