DEV: Added runtimes script

.gitignore

@@ -384,3 +384,7 @@ datasets/AVGFP/Recomb_Double_Split/Recomb_Double_Split4.fasta
 datasets/AVGFP/Recomb_Double_Split/Recomb_Double_Split5.fasta
 datasets/AVGFP/Recomb_Double_Split/Recomb_Double_Split6.fasta
 datasets/AVGFP/Recomb_Double_Split/Recomb_Double_Split7.fasta
+avGFP_shortened_dca_encoded.csv
+datasets/AVGFP/avGFP_shortened.csv
+avGFP_dca_encoded.csv
+scripts/Runtime_tests/runtimes.png

scripts/Runtime_tests/README.md

@@ -1,4 +1,7 @@
-TODO
 ## Benchmarking runtimes
 
-CPU runtimes using multiple threads.
+Wall clock runtimes for PLMC-DCA CPU-based sequence encoding (with PLMC parameter file) using multiple cores/threads.
+
+<p align="center">
+    <img src="../../.github/imgs/runtimes.png" alt="drawing" width="500"/>
+</p>

scripts/Runtime_tests/os_system_runtimes.py

@@ -0,0 +1,44 @@
+
+
+import os
+import sys
+import time
+import matplotlib.pyplot as plt
+
+pypef_path = os.path.abspath(
+    os.path.join(
+        os.path.dirname(__file__),
+        '../..'
+    )
+)
+
+sys.path.append(pypef_path)
+
+avgfp_path = os.path.abspath(
+    os.path.join(
+        os.path.dirname(__file__),
+        '../../datasets/AVGFP'
+    )
+)
+
+# Assuming that the Conda environment 'pypef' exists and contains all necessary Python packages, 
+# using'avGFP_shortened.csv' instead of 'avGFP.csv' takes much less computing time
+cmd = f"conda run -n pypef python {os.path.join(pypef_path, 'pypef', 'main.py')} "\
+      f"encode -i {os.path.join(avgfp_path, 'avGFP.csv')} "\
+      f"-e dca -w {os.path.join(avgfp_path, 'P42212_F64L.fasta')} "\
+      f"--params {os.path.join(avgfp_path, 'uref100_avgfp_jhmmer_119_plmc_42.6.params')} "\
+      f"--threads XX"
+
+print(os.cpu_count())
+all_run_times = []
+for n_cores in range(1, os.cpu_count() + 1):
+    run_time_1 = time.time()
+    print(f"Running command:\n============\n{cmd.replace('XX', str(n_cores))}")
+    os.system(cmd.replace('XX', str(n_cores)))
+    run_time_2 = time.time()
+    all_run_times.append(run_time_2 - run_time_1)
+plt.plot(range(1, os.cpu_count() + 1), all_run_times, 'o--')
+plt.grid()
+plt.xlabel('# Cores/Threads')
+plt.ylabel('Runtime (s)')
+plt.savefig(os.path.join(os.path.dirname(__file__), 'runtimes.png'), dpi=300)