merge local-pc

EleutherAI · May 22, 2021 · eb7ab95 · eb7ab95
2 parents e2e29ea + 4cabbe5
commit eb7ab95
Show file tree

Hide file tree

Showing 14 changed files with 333 additions and 36 deletions.
diff --git a/notebooks/experiments/[131, 150]_info.joblib b/notebooks/experiments/[131, 150]_info.joblib
diff --git a/notebooks/experiments/[200, 250]_info.joblib b/notebooks/experiments/[200, 250]_info.joblib
diff --git a/notebooks/experiments/[331, 351]_info.joblib b/notebooks/experiments/[331, 351]_info.joblib
diff --git a/notebooks/experiments/[400, 450]_info.joblib b/notebooks/experiments/[400, 450]_info.joblib
diff --git a/notebooks/experiments/[500, 550]_info.joblib b/notebooks/experiments/[500, 550]_info.joblib
diff --git a/notebooks/experiments/[600, 650]_info.joblib b/notebooks/experiments/[600, 650]_info.joblib
diff --git a/notebooks/experiments/[700, 780]_info.joblib b/notebooks/experiments/[700, 780]_info.joblib
diff --git a/notebooks/experiments/[800, 900]_info.joblib b/notebooks/experiments/[800, 900]_info.joblib
diff --git a/notebooks/experiments/[905, 1070]_info.joblib b/notebooks/experiments/[905, 1070]_info.joblib
diff --git a/notebooks/experiments/[905, 970]_info.joblib b/notebooks/experiments/[905, 970]_info.joblib
diff --git a/notebooks/experiments/logs_experiment.txt b/notebooks/experiments/logs_experiment.txt
@@ -1,3 +1,4 @@
+<<<<<<< HEAD
 
 2021-05-17 00:00:07,330 INFO MainThread root Loading data
 
@@ -596,3 +597,193 @@
 
 2021-05-17 21:30:24,823 INFO MainThread root Execution has finished
 
+=======
+2021-05-22 02:14:49,435 INFO MainThread root Assessing lengths of: [134, 214, 336, 401, 501, 621, 753, 862, 994]
+
+2021-05-22 02:14:49,435 INFO MainThread root Preparing speed tests: for device 'cpu'
+
+2021-05-22 02:14:49,448 INFO MainThread root Assessing the speed of folding algorithm at length 134
+
+2021-05-22 02:14:55,960 INFO MainThread root 6.509940282999999 for 1000 calls
+2021-05-22 02:14:55,960 INFO MainThread root Saving the related information at experiments/[131, 150]_info.joblib
+
+2021-05-22 02:14:55,961 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:14:55,977 INFO MainThread root Assessing the speed of folding algorithm at length 214
+
+2021-05-22 02:15:02,783 INFO MainThread root 6.805784261999996 for 1000 calls
+2021-05-22 02:15:02,783 INFO MainThread root Saving the related information at experiments/[200, 250]_info.joblib
+
+2021-05-22 02:15:02,784 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:15:02,808 INFO MainThread root Assessing the speed of folding algorithm at length 336
+
+2021-05-22 02:15:11,765 INFO MainThread root 8.956757892999995 for 1000 calls
+2021-05-22 02:15:11,765 INFO MainThread root Saving the related information at experiments/[331, 351]_info.joblib
+
+2021-05-22 02:15:11,766 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:15:11,794 INFO MainThread root Assessing the speed of folding algorithm at length 401
+
+2021-05-22 02:15:21,825 INFO MainThread root 10.031293943000009 for 1000 calls
+2021-05-22 02:15:21,825 INFO MainThread root Saving the related information at experiments/[400, 450]_info.joblib
+
+2021-05-22 02:15:21,826 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:15:21,862 INFO MainThread root Assessing the speed of folding algorithm at length 501
+
+2021-05-22 02:15:33,083 INFO MainThread root 11.221263701000012 for 1000 calls
+2021-05-22 02:15:33,083 INFO MainThread root Saving the related information at experiments/[500, 550]_info.joblib
+
+2021-05-22 02:15:33,084 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:15:33,126 INFO MainThread root Assessing the speed of folding algorithm at length 621
+
+2021-05-22 02:15:45,854 INFO MainThread root 12.727750233999998 for 1000 calls
+2021-05-22 02:15:45,854 INFO MainThread root Saving the related information at experiments/[600, 650]_info.joblib
+
+2021-05-22 02:15:45,855 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:15:45,906 INFO MainThread root Assessing the speed of folding algorithm at length 753
+
+2021-05-22 02:16:00,667 INFO MainThread root 14.760831587000013 for 1000 calls
+2021-05-22 02:16:00,672 INFO MainThread root Saving the related information at experiments/[700, 780]_info.joblib
+
+2021-05-22 02:16:00,674 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:16:00,734 INFO MainThread root Assessing the speed of folding algorithm at length 862
+
+2021-05-22 02:16:17,315 INFO MainThread root 16.580566616 for 1000 calls
+2021-05-22 02:16:17,315 INFO MainThread root Saving the related information at experiments/[800, 900]_info.joblib
+
+2021-05-22 02:16:17,316 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:16:17,383 INFO MainThread root Assessing the speed of folding algorithm at length 994
+
+2021-05-22 02:16:35,654 INFO MainThread root 18.271017204000003 for 1000 calls
+2021-05-22 02:16:35,654 INFO MainThread root Saving the related information at experiments/[905, 1070]_info.joblib
+
+2021-05-22 02:16:35,655 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:16:35,655 INFO MainThread root Preparing speed tests: for device device(type='cpu')
+
+2021-05-22 02:16:35,664 INFO MainThread root Assessing the speed of folding algorithm at length 134
+
+2021-05-22 02:16:40,994 INFO MainThread root 5.329709648000005 for 1000 calls
+2021-05-22 02:16:40,994 INFO MainThread root Saving the related information at experiments/[131, 150]_info.joblib
+
+2021-05-22 02:16:40,995 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:16:41,010 INFO MainThread root Assessing the speed of folding algorithm at length 214
+
+2021-05-22 02:16:47,511 INFO MainThread root 6.501463223999991 for 1000 calls
+2021-05-22 02:16:47,512 INFO MainThread root Saving the related information at experiments/[200, 250]_info.joblib
+
+2021-05-22 02:16:47,513 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:16:47,536 INFO MainThread root Assessing the speed of folding algorithm at length 336
+
+2021-05-22 02:16:56,197 INFO MainThread root 8.660352851999988 for 1000 calls
+2021-05-22 02:16:56,197 INFO MainThread root Saving the related information at experiments/[331, 351]_info.joblib
+
+2021-05-22 02:16:56,198 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:16:56,226 INFO MainThread root Assessing the speed of folding algorithm at length 401
+
+2021-05-22 02:17:05,869 INFO MainThread root 9.643088333000009 for 1000 calls
+2021-05-22 02:17:05,869 INFO MainThread root Saving the related information at experiments/[400, 450]_info.joblib
+
+2021-05-22 02:17:05,871 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:17:05,904 INFO MainThread root Assessing the speed of folding algorithm at length 501
+
+2021-05-22 02:17:17,308 INFO MainThread root 11.40289807900001 for 1000 calls
+2021-05-22 02:17:17,308 INFO MainThread root Saving the related information at experiments/[500, 550]_info.joblib
+
+2021-05-22 02:17:17,309 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:17:17,351 INFO MainThread root Assessing the speed of folding algorithm at length 621
+
+2021-05-22 02:17:30,116 INFO MainThread root 12.764849003000023 for 1000 calls
+2021-05-22 02:17:30,117 INFO MainThread root Saving the related information at experiments/[600, 650]_info.joblib
+
+2021-05-22 02:17:30,119 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:17:30,171 INFO MainThread root Assessing the speed of folding algorithm at length 753
+
+2021-05-22 02:17:44,858 INFO MainThread root 14.687164622000012 for 1000 calls
+2021-05-22 02:17:44,858 INFO MainThread root Saving the related information at experiments/[700, 780]_info.joblib
+
+2021-05-22 02:17:44,859 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:17:44,918 INFO MainThread root Assessing the speed of folding algorithm at length 862
+
+2021-05-22 02:18:01,473 INFO MainThread root 16.554769015000005 for 1000 calls
+2021-05-22 02:18:01,473 INFO MainThread root Saving the related information at experiments/[800, 900]_info.joblib
+
+2021-05-22 02:18:01,474 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:18:01,538 INFO MainThread root Assessing the speed of folding algorithm at length 994
+
+2021-05-22 02:18:19,650 INFO MainThread root 18.111747613000006 for 1000 calls
+2021-05-22 02:18:19,650 INFO MainThread root Saving the related information at experiments/[905, 1070]_info.joblib
+
+2021-05-22 02:18:19,651 INFO MainThread root 
+
+=======
+
+
+2021-05-22 02:18:19,651 INFO MainThread root Execution has finished
+
+>>>>>>> 4cabbe55371d6a9a7edeab1db719fa0cf8312eae
diff --git a/notebooks/experiments_manual/analyzed_prots.joblib b/notebooks/experiments_manual/analyzed_prots.joblib
diff --git a/notebooks/integrated_test.py b/notebooks/integrated_test.py
@@ -1,3 +1,4 @@
+<<<<<<< HEAD
 ##########################
 # Clone repos with utils #
 ##########################
@@ -88,3 +89,108 @@
 
 
 
+=======
+##########################
+# Clone repos with utils #
+##########################
+
+# !git clone https://github.com/hypnopump/geometric-vector-perceptron
+
+import os
+import sys
+import time
+import timeit
+import logging
+
+# science
+import numpy as np 
+import torch
+import sidechainnet
+from sidechainnet.utils.sequence import ProteinVocabulary as VOCAB
+VOCAB = VOCAB()
+
+# process
+import joblib
+
+# custom
+import mp_nerf
+
+BASE_FOLDER = "experiments/"
+
+logging.basicConfig(level=logging.DEBUG,
+                    format="%(asctime)s %(levelname)s %(threadName)s %(name)s %(message)s",
+                    # datefmt='%m-%d %H:%M',
+                    filename=BASE_FOLDER+"logs_experiment.txt",
+                    filemode="a")
+logger = logging.getLogger()
+sep = "\n\n=======\n\n"
+
+
+# begin tests
+if __name__ == "__main__":
+    "a"+9
+    logger.info("Loading data"+"\n")
+    # adapt desired lengths to previous experiments results
+    # lengths = [100, 200, 300, 400, 500, 600, 700, 800, 900]# [::-1]
+    lengths = [[131, 150],
+               [200, 250],
+               [331, 351],
+               [400, 450],
+               [500, 550],
+               [600, 650],
+               [700, 780],
+               [800, 900],
+               [905, 1070]]# [::-1]
+    try: 
+        # skip
+        logger.info("Data has been loaded"+"\n"+sep)
+        stored  = [ ]
+        for i,(desired_len, upper_len) in enumerate(lengths[::-1]): 
+            dataloaders_ = sidechainnet.load(casp_version=7, with_pytorch="dataloaders", batch_size=2)
+            stored.append(mp_nerf.utils.get_prot(dataloader_=dataloaders_, 
+                                                 vocab_=VOCAB, 
+                                                 min_len=desired_len, 
+                                                 max_len=upper_len))
+        stored = stored[::-1]
+
+        joblib.dump(stored, BASE_FOLDER[:-1]+"_manual/analyzed_prots.joblib")
+    except: 
+        stored = joblib.load(BASE_FOLDER[:-1]+"_manual/analyzed_prots.joblib")
+        logger.info("Data has been loaded"+"\n"+sep)
+
+    logger.info("Assessing lengths of: "+str([len(x[0]) for x in stored])+"\n")
+
+    cuda_device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    for device in ["cpu", cuda_device]:
+
+        logger.info("Preparing speed tests: for device "+repr(device)+"\n")
+
+        for i,desired_len in enumerate(lengths):
+
+            seq, int_seq, true_coords, angles, padding_seq, mask, pid = stored[i]
+            scaffolds = mp_nerf.proteins.build_scaffolds_from_scn_angles(seq, angles.to(device))
+
+            logger.info("Assessing the speed of folding algorithm at length "+str(len(seq))+"\n")
+
+            logger.info( str( timeit.timeit('mp_nerf.proteins.protein_fold(**scaffolds, device=device, hybrid=True)',
+            	                             globals=globals(), number=1000) )+" for 1000 calls" )
+
+            logger.info("Saving the related information at {0}{1}_info.joblib\n".format(
+                        BASE_FOLDER, desired_len))
+            joblib.dump({"seq": seq, 
+                         "true_coords": true_coords,
+                         "angles": angles,
+                         "padding_seq": padding_seq,
+                         "mask": mask,
+                         "pid": pid, 
+                         "padding_stripped": True}, BASE_FOLDER+str(desired_len)+"_info.joblib")
+            logger.info(sep)
+
+    logger.info("Execution has finished\n")
+
+
+
+
+
+
+>>>>>>> 4cabbe55371d6a9a7edeab1db719fa0cf8312eae
diff --git a/setup.py b/setup.py
@@ -1,36 +1,36 @@
-from setuptools import setup, find_packages
-
-setup(
-  name = 'mp-nerf',
-  packages = find_packages(),
-  version = '0.1.7',
-  license='MIT',
-  description = 'MP-NeRF: Massively Parallel Natural Extension of Reference Frame',
-  author = 'Eric Alcaide',
-  author_email = '[email protected]',
-  url = 'https://github.com/eleutherAI/mp_nerf',
-  keywords = [
-    'computational biolgy',
-    'bioinformatics',
-    'machine learning' 
-  ],
-  install_requires=[
-    'einops>=0.3',
-    'numpy',
-    'torch>=1.6',
-    'sidechainnet'
-  ],
-  setup_requires=[
-    'pytest-runner',
-  ],
-  tests_require=[
-    'pytest'
-  ],
-  classifiers=[
-    'Development Status :: 4 - Beta',
-    'Intended Audience :: Developers',
-    'Topic :: Scientific/Engineering :: Artificial Intelligence',
-    'License :: OSI Approved :: MIT License',
-    'Programming Language :: Python :: 3.6',
-  ],
-)
+from setuptools import setup, find_packages
+
+setup(
+  name = 'mp-nerf',
+  packages = find_packages(),
+  version = '0.1.7',
+  license='MIT',
+  description = 'MP-NeRF: Massively Parallel Natural Extension of Reference Frame',
+  author = 'Eric Alcaide',
+  author_email = '[email protected]',
+  url = 'https://github.com/eleutherAI/mp_nerf',
+  keywords = [
+    'computational biolgy',
+    'bioinformatics',
+    'machine learning' 
+  ],
+  install_requires=[
+    'einops>=0.3',
+    'numpy',
+    'torch>=1.6',
+    'sidechainnet'
+  ],
+  setup_requires=[
+    'pytest-runner',
+  ],
+  tests_require=[
+    'pytest'
+  ],
+  classifiers=[
+    'Development Status :: 4 - Beta',
+    'Intended Audience :: Developers',
+    'Topic :: Scientific/Engineering :: Artificial Intelligence',
+    'License :: OSI Approved :: MIT License',
+    'Programming Language :: Python :: 3.6',
+  ],
+)