From 900b199b4f97b348d49fb7f134c58e57cde3f98d Mon Sep 17 00:00:00 2001
From: Nico Maillet <nicolas.maillet@pasteur.fr>
Date: Wed, 20 Sep 2023 10:43:18 +0200
Subject: [PATCH] A tp2 soluce
---
2-Data-structures_1/tp2.py | 125 +++++++++++++++++++++++++++++++++++++
1 file changed, 125 insertions(+)
create mode 100644 2-Data-structures_1/tp2.py
diff --git a/2-Data-structures_1/tp2.py b/2-Data-structures_1/tp2.py
new file mode 100644
index 0000000..71efe35
--- /dev/null
+++ b/2-Data-structures_1/tp2.py
@@ -0,0 +1,125 @@
+""" TP2 where we code some reverse complement functions """
+# Import random and time
+import random
+import time
+
+def reverse_complement_naive(seq):
+ """ Take an input sequence and return its revcomp """
+ # Empty return sequence
+ ret = ""
+ # For each nucleotide in seq
+ for nucl in seq:
+ # If it is an A
+ if nucl == "A":
+ # add T at the beginning of ret
+ ret = "T" + ret
+ # If it is an T
+ if nucl == "T":
+ # add A at the beginning of ret
+ ret = "A" + ret
+ # If it is an C
+ if nucl == "C":
+ # add G at the beginning of ret
+ ret = "G" + ret
+ # If it is an G
+ if nucl == "G":
+ # add C at the beginning of ret
+ ret = "C" + ret
+ # Return ret
+ return ret
+
+def reverse_complement_naive2(seq):
+ """ Take an input sequence and return its revcomp """
+ # Empty return sequence
+ ret = ""
+ # For each nucleotide IN REVERSE SEQ
+ for nucl in seq[::-1]:
+ # If it is an A
+ if nucl == "A":
+ # add T
+ ret += "T"
+ # If it is an T
+ if nucl == "T":
+ # add A
+ ret += "A"
+ # If it is an C
+ if nucl == "C":
+ # add G
+ ret += "G"
+ # If it is an G
+ if nucl == "G":
+ # add C
+ ret += "C"
+ # Return ret
+ return ret
+
+def reverse_complement_less_naive(seq):
+ """ Take an input sequence and return its revcomp """
+ # Hash of correspondence
+ rv_val = {"A": "T", "T": "A", "C": "G", "G": "C"}
+ # Empty return sequence
+ ret = ""
+ # For each nucleotide
+ for nucl in seq[::-1]:
+ # Add its rv to ret
+ ret += rv_val[nucl]
+ # Return ret
+ return ret
+
+def generator_of_sequence(size):
+ """ Generate a random ntd sequence of length 'size' """
+ # Return sequence
+ ret = ""
+ # While we did not have a long enough sequence
+ for _ in range(size):
+ # Add a new random ntd
+ ret += random.choice("ACGT")
+ # Return the sequence
+ return ret
+
+def main():
+ """ The main of TP2 that compare different revcomp functions """
+
+ # Create a random sequence of specific size
+ seq = generator_of_sequence(1000000)
+
+ # Get the current time
+ start = time.time()
+ # Test the first naive revcomp function
+ res_naive = reverse_complement_naive(seq)
+ # Get the current time
+ end = time.time()
+ # Print the execution time (2 digits precision)
+ print(f"Naive: {end - start:.2f}s")
+
+ # Get the current time
+ start = time.time()
+ # Test the second naive revcomp function
+ res_naive2 = reverse_complement_naive2(seq)
+ # Get the current time
+ end = time.time()
+ # Print the execution time (2 digits precision)
+ print(f"Naive2: {end - start:.2f}s")
+
+ # Get the current time
+ start = time.time()
+ # Test the less naive revcomp function
+ res_less_naive = reverse_complement_less_naive(seq)
+ # Get the current time
+ end = time.time()
+ # Print the execution time (2 digits precision)
+ print(f"Less naive: {end - start:.2f}s")
+
+ # Ensure that all results are identical
+ if res_naive == res_naive2 and res_naive == res_less_naive:
+ # Print that everything is good
+ print("Sequences are identical")
+ else:
+ # Print that there is a problem somewhere
+ print("There is a bug!!")
+
+# Launch the main
+main()
+# Exit without error
+exit(0)
+# Always put one extra return line
--
GitLab