Merge pull request #44 from PoonLab/dev

fixes to python interface

Author: ArtPoon

bin/openrdp

@@ -15,7 +15,8 @@ if __name__ == "__main__":
     )
     parser.add_argument('infile',
                         help='File containing sequence alignment (FASTA or CLUSTAL) format')
-    parser.add_argument('outfile', help='Path to the output file')
+    parser.add_argument('-o', '--outfile', type=argparse.FileType('w'),
+                        default=None, help='Path to write CSV output')
     parser.add_argument('-c', '--cfg', type=str,
                         help='Path to file that contains parameters.  Defaults to '
                              'default.ini.',
@@ -36,7 +37,8 @@ if __name__ == "__main__":
     if args.methods is None:
         args.methods = all_methods  # run all methods by default
 
-    results = openrdp.openrdp(args.infile, args.outfile, cfg=args.cfg,
-                              methods=tuple(args.methods), quiet=args.quiet)
+    results = openrdp.openrdp(args.infile, cfg=args.cfg, methods=tuple(args.methods),
+                              quiet=args.quiet)
     print(results)  # calls custom __str__
-
+    if args.outfile:
+        results.write(args.outfile)

openrdp/__init__.py

@@ -5,6 +5,7 @@
 from itertools import combinations
 from glob import glob
 
+from openrdp import __path__ as basepath
 from openrdp.bootscan import Bootscan
 from openrdp.chimaera import Chimaera
 from openrdp.common import generate_triplets, Triplet
@@ -63,32 +64,41 @@ def __str__(self):
                               f"{e[1][0]:<7}\t{e[1][1]:<7}\t{float(e[4]):.2E}\n"
         return outstr
 
+    def __getitem__(self, key):
+        events = self.dict.get(key, None)
+        if key == 'geneconv':
+            return [{'start': e[2][0], 'end': e[2][1], 'recombinant': e[0],
+                     'parent1': e[1][0], 'parent2': e[1][1], 'pvalue': float(e[3])}
+                    for e in events]
+        else:
+            return [{'start': e[2], 'end': e[3], 'recombinant': e[0],
+                     'parent1': e[1][0], 'parent2': e[1][1], 'pvalue': float(e[4])}
+                    for e in events]
+
+    def keys(self):
+        return self.dict.keys()
+
 
 class Scanner:
-    def __init__(self, names, infile, outfile, cfg=None, methods=None,
+    def __init__(self, names, infile, cfg=None, methods=None,
                  quiet=False):
         """
         :param names:  list, sequence labels
         :param infile:  str, path to input FASTA
-        :param outfile:  str, path to write output CSV
         :param cfg:  str, path to configuration file.  Defaults to None, causing
                      each method to use default settings.
         :param methods:  tuple, names of methods to run
         :param quiet:  bool, if True, suppress console messages
         """
         self.seq_names = names
         self.infile = infile
-
-        self.config = None
-        self.cfg_file = cfg
-        if self.cfg_file:
-            print(f"Loading configuration from {cfg}")
-            self.config = configparser.ConfigParser()
-            self.config.read(self.cfg_file)
-
         self.methods = methods
         self.quiet = quiet
-        self.outfile = outfile
+
+        self.config = configparser.ConfigParser()
+        self.cfg_file = cfg
+        self.print(f"Loading configuration from {self.cfg_file}")
+        self.config.read(self.cfg_file)
 
     def print(self, msg):
         """ Implements self.quiet """
@@ -277,11 +287,10 @@ def read_fasta(handle):
     return headers, seqs
 
 
-def openrdp(infile, outfile, cfg=None, methods=None, quiet=False):
+def openrdp(infile, cfg=None, methods=None, quiet=True):
     """
     Main function
     :param infile:  str, path to input FASTA file
-    :param outfile:  str, path to write output CSV
     :param methods:  list, names of methods to run
     :param quiet:  bool, if True, suppress console messages
     :return:  dict, results from each method
@@ -306,6 +315,10 @@ def openrdp(infile, outfile, cfg=None, methods=None, quiet=False):
     if not valid_alignment(aln) and not valid_chars(aln):
         sys.exit(1)
 
-    scanner = Scanner(names, infile, outfile, cfg=cfg, methods=methods, quiet=quiet)
+    if cfg is None:
+        # load default configuration from package file
+        cfg = os.path.join(basepath[0], 'default.ini')
+
+    scanner = Scanner(names, infile, cfg=cfg, methods=methods, quiet=quiet)
     results = scanner.run_scans(aln)
     return results

openrdp/bootscan.py

@@ -31,12 +31,13 @@ def __init__(self, alignment, win_size=200, step_size=20, use_distances=True,
         self.align = alignment
         random.seed(self.random_seed)
         np.random.seed(self.random_seed)
-        if not quiet:
+
+        self.quiet = quiet
+        if not self.quiet:
             print('Starting Scanning Phase of Bootscan/Recscan')
 
         self.do_scanning_phase(alignment)
-
-        if not quiet:
+        if not self.quiet:
             print('Finished Scanning Phase of Bootscan/Recscan')
 
         self.raw_results = []
@@ -166,7 +167,8 @@ def execute(self, arg):
         raw_results = []
         triplet = Triplet(self.align, self.seq_names, trp)
 
-        print("Scanning triplet {} / {}".format(i, self.total_triplet_combinations))
+        if not self.quiet:
+            print("Scanning triplet {} / {}".format(i, self.total_triplet_combinations))
 
         # Look at boostrap support for sequence pairs
         ab_support = [0] * (self.align.shape[1] // self.step_size)