Merge branch 'fix_logging' into 'main'

feat: added random sequence generator to cli

See merge request !21

read_sequencer_package/cli.py

 import argparse
-from modules import read_sequencer as rs
+from modules import ReadSequencer
 import logging
 
 parser = argparse.ArgumentParser(prog='read_sequencer',
@@ -11,21 +11,28 @@ parser.add_argument('--output_file_path',
 parser.add_argument('--read_length',
                     help='read length for sequencing',
                     type=int)
+parser.add_argument('--random', action='store_true', default=False,
+                    help='generate random sequences')
+parser.add_argument('--n_random', default=100, type=int, help='n random sequences')
+parser.add_argument('--mean_random', default=50, type=int, help='mean random sequences')
+parser.add_argument('--sd_random', default=25, type=int, help='standard deviation random sequences')
 
 args = parser.parse_args()
 
 def main():
-    LOG.info("Program started.")
-    read_sequencer = rs()
-    read_sequencer.read_fasta(args.input_file_path)
+    LOG.info("Read sequencer started.")
+    read_sequencer = ReadSequencer()
+    if args.random:
+        read_sequencer.add_random_sequences(n=args.n_random, mean=args.mean_random, sd=args.sd_random)
+    else:
+        read_sequencer.read_fasta(args.input_file_path)
     read_sequencer.run_sequencing(args.read_length)
     read_sequencer.write_fasta(args.output_file_path)
-    LOG.info("Program finished.")
+    LOG.info("Read sequencer finished.")
 
 if __name__ == '__main__':
     logging.basicConfig(
         format='[%(asctime)s: %(levelname)s] %(message)s (module "%(module)s")',
-        level=logging.INFO,
-    )
+        level=logging.INFO)
     LOG = logging.getLogger(__name__)
     main()

read_sequencer_package/modules.py

 import logging
 LOG = logging.getLogger(__name__)
 
-def generate_sequences(n, mean, sd):
-    """
-    Generates random sequences.
-
-    Args:
-        n (int): Amount of sequences to generate.
-        mean (int): mean length of sequence (gaussian distribution).
-        sd (float): standard deviation of length of sequence (gaussian distribution).
-
-    Returns:
-        list: of n sequences
-    """
-    from random import gauss, choice
-    LOG.info("Generating sequences.")
-    dict = {}
-    for i in range(n):
-        keys = range(n)
-        seq = ""
-        nt = ["A", "T", "C", "G"]
-        for value in range(abs(round(gauss(mean, sd)))):
-            seq = seq + choice(nt)
-        dict[keys[i]] = seq
-    return dict
-
-
 def read_in_fasta(file_path):
-    '''
+    """
     This function reads in FASTA files.
 
     Args:
@@ -36,7 +11,7 @@ def read_in_fasta(file_path):
     Returns:
         Dict: It returns a dictionary with sequences.
 
-    '''
+    """
     LOG.info("Reading in FASTA files from destination.")
     sequences = {}
     f = open(file_path)
@@ -52,7 +27,7 @@ def read_in_fasta(file_path):
     return sequences
 
 def read_sequence(seq, read_length):
-    '''
+    """
     This function reads a sequence of a specific read length and adds additional nucleotides if the sequence is 
     smaller then the requested length or cuts the sequence if its longer.
 
@@ -63,12 +38,11 @@ def read_sequence(seq, read_length):
     Returns:
         str: returns sequenced element
 
-    '''
-    
+    """
     from random import choice
     bases = ["A", "T", "C", "G"]
     sequenced = ''
-    if read_length >= len(seq):
+    if read_length > len(seq):
         for nt in range(len(seq)):
             sequenced += seq[nt]
         for nt in range(len(seq), read_length):
@@ -94,10 +68,10 @@ def simulate_sequencing(sequences, read_length):
     results = {}
     for index, key in enumerate(sequences):
         results[key] = read_sequence(sequences[key], read_length=read_length)
-
+    LOG.info("Sequencing was successfully executed.")
     return results
 
-import random
+
 def generate_sequences(n, mean, sd):
     """
     Generates random sequences.
@@ -110,16 +84,17 @@ def generate_sequences(n, mean, sd):
     Returns:
         dict: of n sequences
     """
+    from random import choice, gauss
     LOG.info("Generating random sequences.")
-    dict1 = {}
+    dict = {}
     for i in range(n):
-        keys = range(n)
         seq = ""
         nt = ["A", "T", "C", "G"]
-        for value in range(round(random.gauss(mean, sd))):
-            seq = seq + random.choice(nt)
-        dict1[keys[i]] = seq
-    return dict1
+        for value in range(abs(round(gauss(mean, sd)))):
+            seq = seq + choice(nt)
+        key = str(i) + ': length ' + str(len(seq)) + ' nt'
+        dict[key] = seq
+    return dict
 
 def write_fasta(sequences, file_path):
     """
@@ -138,8 +113,8 @@ def write_fasta(sequences, file_path):
             outfile.write(key + "\n")
             outfile.write("\n".join(wrap(value, 60)))
             outfile.write("\n")
-    LOG.info("Sequencing was successfully executed.")
-class read_sequencer:
+
+class ReadSequencer:
     def __init__(self):
         self.sequences = {}
         self.reads = {}