Merge pull request #41 from jvalegre/ddg_branch

1. Update v1.1.1
jvalegre · Jul 31, 2024 · f80ede8 · f80ede8
2 parents 703d589 + 87f5909
commit f80ede8
Show file tree

Hide file tree

Showing 6 changed files with 64 additions and 46 deletions.
diff --git a/docs/Examples/full_workflow/smiles_vaskas.rst b/docs/Examples/full_workflow/smiles_vaskas.rst
@@ -17,8 +17,9 @@ Reproducibility warning
 
 .. warning::
 
-   **Update to AQME v1.6.0 or higher to obtain fully reproducible results!** Otherwise, it may not be possible to exactly 
-   reproduce the results due to subtle differences in the generated xTB descriptors (0.1% changes in most cases).
+   **Update to AQME v1.6.0 or higher to obtain fully reproducible results!** You can do it with :code:`pip install aqme --upgrade`.
+   Otherwise, it may not be possible to exactly reproduce the results due to subtle differences in the 
+   generated xTB descriptors (0.1% changes in most cases).
 
 Required inputs
 +++++++++++++++

diff --git a/docs/Examples/full_workflow/smiles_workflow.rst b/docs/Examples/full_workflow/smiles_workflow.rst
@@ -14,8 +14,9 @@ Reproducibility warning
 
 .. warning::
 
-   **Update to AQME v1.6.0 or higher to obtain fully reproducible results!** Otherwise, it may not be possible to exactly 
-   reproduce the results due to subtle differences in the generated xTB descriptors (0.1% changes in most cases).
+   **Update to AQME v1.6.0 or higher to obtain fully reproducible results!** You can do it with :code:`pip install aqme --upgrade`.
+   Otherwise, it may not be possible to exactly reproduce the results due to subtle differences in the 
+   generated xTB descriptors (0.1% changes in most cases).
 
 Required inputs
 +++++++++++++++

diff --git a/docs/README.rst b/docs/README.rst
@@ -113,17 +113,24 @@ it also works for other Python versions (i.e., 3.7, 3.9 and 3.11):
    conda install -c conda-forge robert
    pip install scikit-learn-intelex
 
-**3.** Install GLib to avoid errors when creating the PDF report:  
+**3.** Update ROBERT to the latest version (do not skip this step!):
+
+.. code-block:: shell 
+
+   pip install robert --upgrade
+
+**4.** Install GLib and mscorefonts to avoid errors when creating the PDF report:  
 
 .. code-block:: shell 
    
    conda install -c conda-forge glib
+   conda install -c conda-forge mscorefonts
 
 .. warning::
 
    In some computers, you might see lots of pop-up windows with errors when installing GLib. Just press "Accept" in all the windows and finish the process until you see "Done" in the terminal.
 
-**4.** If the installation of steps 1-3 with conda-forge are too slow, users might install ROBERT using pip instead. Then, install the libraries required for report.py and the intelex accelerator (only if your system is compatible with intelex):  
+**(Alternative installation).** If the installation of steps 2-4 with conda-forge are too slow, users might install ROBERT using pip instead. Then, install the libraries required for report.py and the intelex accelerator (only if your system is compatible with intelex):  
 
 .. code-block:: shell
 
@@ -173,10 +180,13 @@ you can install `Miniconda with Python 3 <https://docs.conda.io/projects/minicon
 **7.** Update ROBERT to the latest version (do not skip this step!) (:code:`pip install robert --upgrade`).
 
 
-**8.** Go to the folder with your CSV database (using the "cd" command, i.e. :code:`cd C:/Users/test_robert`).
+**8.** Install GLib and mscorefonts to avoid errors when creating the PDF report (:code:`conda install -c conda-forge glib` and :code:`conda install -c conda-forge mscorefonts`).
+
+
+**9.** Go to the folder with your CSV database (using the "cd" command, i.e. :code:`cd C:/Users/test_robert`).
 
 
-**9.** Run ROBERT as explained in the Examples section.
+**10.** Run ROBERT as explained in the Examples section.
 
 .. note-end 
 

diff --git a/robert/report.py b/robert/report.py
@@ -411,7 +411,7 @@ def get_repro(self,modules):
         citation_dat += f"""<p style="text-align: justify;"><br>{version_n_date}</p>
         <p style="text-align: justify;  margin-top: -8px;"><span style="font-weight:bold;">How to cite:</span> {citation}</p>"""
 
-        aqme_workflow = False
+        aqme_workflow,aqme_updated = False,True
         crest_workflow = False
         if '--aqme' in command_line:
             original_command = command_line
@@ -442,30 +442,43 @@ def get_repro(self,modules):
             repro_dat += f"""{reduced_line}{space}- External test set ({self.args.csv_test})</p>"""
 
         if aqme_workflow:
-            repro_dat += f"""{reduced_line}{space}<i>Warning! SMILES workflows might not be reproduced exactly due to subtle changes in xTB descriptors (&plusmn;0.1%):</i></p>"""
-            repro_dat += f"""{reduced_line}{space}To obtain the same results, download the descriptor database (AQME-ROBERT_{self.args.csv_name}) and run:</p>"""
-            repro_line = []
-            original_command = original_command.replace(self.args.csv_name,f'AQME-ROBERT_{self.args.csv_name}')
-            for i,keyword in enumerate(original_command.split('"')):
-                if i == 0:
-                    if '--aqme' not in keyword and '--qdescp_keywords' not in keyword and '--csearch_keywords' not in keyword:
-                            repro_line.append(keyword)
-                    else:
-                        repro_line.append('python -m robert ')
-                if i > 0:
-                    if '--qdescp_keywords' not in original_command.split('"')[i-1] and '--csearch_keywords' not in original_command.split('"')[i-1]:
-                        if '--aqme' not in keyword and '--qdescp_keywords' not in keyword and '--csearch_keywords' not in keyword and keyword != '\n':
-                            repro_line.append(keyword) 
-            repro_line = '"'.join(repro_line)
-            repro_line += '"'
-            if '--names ' not in repro_line:
-                repro_line += ' --names "code_name"'
-            repro_line = f'{reduced_line}{space}- Run: {repro_line}'
-            repro_line = format_lines(repro_line,cmd_line=True)
-            repro_dat += f"""{reduced_line}{repro_line}</p>"""
-
-        if aqme_workflow:
+            try:
+                path_aqme = Path(f'{os.getcwd()}/AQME/CSEARCH_data.dat')
+                datfile = open(path_aqme, 'r', errors="replace")
+                outlines = datfile.readlines()
+                aqme_version = outlines[0].split()[2]
+                datfile.close()
+                find_aqme = True
+            except:
+                find_aqme = False
+                aqme_version = '0.0' # dummy number
+            if int(aqme_version.split('.')[0]) in [0,1] and int(aqme_version.split('.')[1]) < 6:
+                aqme_updated = False
+                repro_dat += f"""{reduced_line}{space}<i>Warning! This workflow might not be exactly reproducible, update to AQME v1.6.0+ (pip install aqme --upgrade)</i></p>"""
+                repro_dat += f"""{reduced_line}{space}To obtain the same results, download the descriptor database (AQME-ROBERT_{self.args.csv_name}) and run:</p>"""
+                repro_line = []
+                original_command = original_command.replace(self.args.csv_name,f'AQME-ROBERT_{self.args.csv_name}')
+                for i,keyword in enumerate(original_command.split('"')):
+                    if i == 0:
+                        if '--aqme' not in keyword and '--qdescp_keywords' not in keyword and '--csearch_keywords' not in keyword:
+                                repro_line.append(keyword)
+                        else:
+                            repro_line.append('python -m robert ')
+                    if i > 0:
+                        if '--qdescp_keywords' not in original_command.split('"')[i-1] and '--csearch_keywords' not in original_command.split('"')[i-1]:
+                            if '--aqme' not in keyword and '--qdescp_keywords' not in keyword and '--csearch_keywords' not in keyword and keyword != '\n':
+                                repro_line.append(keyword) 
+                repro_line = '"'.join(repro_line)
+                repro_line += '"'
+                if '--names ' not in repro_line:
+                    repro_line += ' --names "code_name"'
+                repro_line = f'{reduced_line}{space}- Run: {repro_line}'
+                repro_line = format_lines(repro_line,cmd_line=True)
+                repro_dat += f"""{reduced_line}{repro_line}</p>"""
+
+        if aqme_workflow and not aqme_updated:
             # I use a very reduced line in this title because the formatted command_line comes with an extra blank line
+            # (if AQME is not updated the PDF contains a reproducibility warning)
             repro_dat += f"""<p style="text-align: justify; margin-top: -44px;"><br><strong>2. Install and adjust the versions of the following Python modules:</strong></p>"""
         else:
             repro_dat += f"""{first_line}<br><strong>2. Install and adjust the versions of the following Python modules:</strong></p>"""
@@ -480,15 +493,6 @@ def get_repro(self,modules):
             repro_dat += f"""{reduced_line}{space}<i>(if scikit-learn-intelex is installed, slightly different results might be obtained)</i></p>"""
 
         if aqme_workflow:
-            try:
-                path_aqme = Path(f'{os.getcwd()}/AQME/CSEARCH_data.dat')
-                datfile = open(path_aqme, 'r', errors="replace")
-                outlines = datfile.readlines()
-                aqme_version = outlines[0].split()[2]
-                datfile.close()
-                find_aqme = True
-            except:
-                find_aqme = False
             if not find_aqme:
                 repro_dat += f"""{reduced_line}{space}- AQME is required, but no version was found:</p>"""
             repro_dat += f"""{reduced_line}{space}- Install AQME and its dependencies: conda install -c conda-forge aqme</p>"""

diff --git a/robert/utils.py b/robert/utils.py
@@ -42,7 +42,7 @@
 import warnings # this avoids warnings from sklearn
 warnings.filterwarnings("ignore")
 
-robert_version = "1.1.0"
+robert_version = "1.1.1"
 time_run = time.strftime("%Y/%m/%d %H:%M:%S", time.localtime())
 robert_ref = "Dalmau, D.; Alegre Requena, J. V. ChemRxiv, 2023, DOI: 10.26434/chemrxiv-2023-k994h"
 
@@ -1047,6 +1047,8 @@ def calc_ci_n_sd(self,loaded_model,data,set_mapie):
     # Check if 1/alpha is lower than the number of samples
     if 1 / self.args.alpha >= len(data[f'X_{set_mapie}_scaled']):
         self.args.alpha = 0.1
+        if 1 / self.args.alpha >= len(data[f'X_{set_mapie}_scaled']):
+            self.args.alpha = 0.5
 
     # Predict test set and obtain prediction intervals
     y_pred, y_pis = mapie.predict(data[f'X_{set_mapie}_scaled'], alpha=[self.args.alpha])

diff --git a/setup.py b/setup.py
@@ -1,5 +1,5 @@
 from setuptools import setup, find_packages
-version="1.1.0"
+version="1.1.1"
 setup(
     name="robert",
     packages=find_packages(exclude=["tests"]),
@@ -33,16 +33,16 @@
     ],
     install_requires=[
         "PyYAML",
+        "matplotlib>=3.7,<3.8",
         "pandas==2.2.2",
-        "progress",
         "numpy==1.26.4",
-        "matplotlib>=3.7,<3.8",
+        "progress",
         "seaborn",
         "scipy",
         "scikit-learn==1.5.1", # fixed to make ROBERT reproducible when using the same version
         "hyperopt",
         "numba==0.60.0",
-        "shap==0.46.0",
+        "shap==0.45.1",
         "mapie==0.8.6"
         # requires also "conda install -c conda-forge gtk3 glib weasyprint pango" for report.py
         # optionally, the intelex accelerator might be installed "pip install scikit-learn-intelex"