.. _versions: ======== Versions ======== Version 2.0.1 [`url `__] - Fixed PermissionError issue in Windows related to Chem.SDMolSupplier() Version 2.0.0 [`url `__] - Amount of generated conformers (--sample) can be defined in the CSV input as a column to set individual --sample values for each molecule - Multiple geometry filters allow through the --geom option (format: --geom [SMARTS1, VALUE1, SMARTS2, VALUE2...]) - Included new option --single_system for CSEARCH to avoid duplicate geometries when using --complex_type templates in complexes with 2 identical ligands - Debugging multithreading to avoid QDESCP problems (includes a forced wait for all the calculations to finish before getting RDKit and atomic properties) - The SUMM and FullMonte options of CSEARCH have been disabled - Big refactoring of the code - cclib updated version to 1.8.1 and called through cclib object (instead ob ccwrite) - RDKit updated version to 2025 - numpy updated to 2.3.4 - Update of QDESCP to go through MORFEUS 0.8.0 Version 1.7.3 [`url `__] - Fixed bug from QDESCP when trying to open JSON files - Using libgfortran=14.2.0 to avoid issues when using xTB and CREST Version 1.7.2 [`url `__] - Matching Pandas and Python versions in setup.py for ALMOS installation Version 1.7.1 [`url `__] - Fixed bug from QCORR when using gen(ECP) - CSEARCH uses multithreading to ensure reproducbility acorss runs - Added T1-S0 gap (not only S0-T1) - PTB is now used to calculate HOMO-LUMO gaps, charges, dipole moments, and other descriptors - Using xTB v6.7.1 and CREST v2.12 - Fixed bug from MORFEUS descriptors in QDESCP - Improved SDF reader compatible with SDF files from GaussView - Refactoring of the QDESCP module and its utils (faster code execution) Version 1.7.0 [`url `__] - Fixed bug from CSEARCH when using CSV inputs with metal atoms (removed SELF variables in multithreading) - Added --sample to CSEARCH-CREST through RMS-based Butina sampling - Changing to pip install by default in Read the Docs installation instructions - CMIN and QDESCP don't stop if an xTB calculation fails, instead they keep running the remaining calculations - CSEARCH doesn't stop if a conformer search fails, instead it keeps running the remaining searches - Adding many new descriptors in QDESCP within three levels of human interpretation (denovo, interpret and full) - Empty generated SDF files are removed - Prints of missing dependencies are also added when using Jupyter Notebooks (not only in the command line) - DBSTEP functions were removed - MORFEUS functions were added to include steric descriptors - Number of initial samples generated by RDKit was limited to 500 - Selection of conformers from RDKit through the sample option is now controlled by a combination of stable RDKit conformers and Butina sampling - Add solvent descriptors in QDESCP - Lists with atom descriptors are not shown in the main descriptor CSV in QDESCP (but they are still stored in the raw_csv_databases folder) - Great acceleration of QDESCP using multithreading across molecules (instead of conformers) - QDESCP generates only 2 raw xTB files (instead of 11), and performs a clean-up before each calculation to avoid reading from previous restart points - qdescp_atoms works from SDF with no property SMILES - Automatic pattern recognition for qdescp_atoms now works for all types of atoms and multiple instances (not only metals) - QDESCP refactoring to make the workflows easier to follow - Add automated workflow from SMILES to descriptors using --input FILENAME.csv in QDESCP - QDESCP calculates charges from SDF that doesn't come from CSEARCH Version 1.6.1 [`url `__] - Hotfix of v1.6.0 (just moving the import of dbstep in QDESCP) Version 1.6.0 [`url `__] - CSEARCH uses multithreading to accelerate jobs - QDESP uses multithreading to accelerate jobs - QDESCP now accepts multi-SMILES columns in CSV files - CSEARCH now handles multiple metals without program interruption - Check for N@@ or N@ in SMILES to avoid RDKit issues - CSEARCH now performs conformational search only for unique SMILES to avoid redundant calculations - QDESCP now automatically obtains SMARTS patterns from input files if no patterns are provided - QDESCP now uses a single processor for xtb calculations to ensure reproducibility - Fixed bug in QCORR when analyzing TD-DFT, ONIOM and G4 calculations in Gaussian - CSEARCH is now compatible with mapped atoms - QPREP now warns if the level of theory is not defined in ORCA Version 1.5.2 [`url `__] - Fixed bug for using constraints with large molecules in CREST (related to long atom lists in the .xcontrol.sample file that weren't compatible with subprocess.run()) - Fixed bug when using squareplanar templates with ligands that bond with the metal centers with the same type of atom (but they are different ligands) - Added the 'Ir_squareplanar' geometry rule - Better print for command lines (more reproducible commands) - QDESCP adds an initial xTB optimization before the single-points (before, the descriptors were obtained from RDKit geometries, and they were pretty bad for metal complexes) - The aqme.py file now takes all the arguments from command lines with vars(args) without specifying the argument names explicitly in the code - Added the chk_path option to add PATH into chk lines - Added the xtb_opt option to optionally disable the initial xTB geometry optimization in QDESCP - Fixed a bug when an xTB calc failed, the whole QDESCP workflow stopped (now the specific conformer just gets ignored) - Fixed bug when GB or MB were not part of the mem argument in QPREP-ORCA - Added the qdescp_opt option for adjusting the convergence criteria of xTB in QDESCP - DBSTEP calcs are deactivated by default in QDESCP (activate them with --dbstep_calc) - --auto_sample now modulates the amount of initial RDKit conformers generated with three options: low, mid and high Version 1.5.1 [`url `__] - Added the verbose option. When verbose=False, no DAT and CSV files with summaries are printed - Fixed bug for using constraints with SMILES that are not mapped - Changed directory to create AQME-ROBERT databases to working dir (to adapt for full ML workflows in ROBERT) - Added SMARTS pattern recogniton to qdescp_atoms (the option now takes lists of atoms as well as lists of functional groups) - Fix minor bugs in QCORR-ORCA - Added the geometry filter (geom option) to CSEARCH - Added geom and complex_type as options in CSV columns - Fixed QDESCP when using csv_name with columns that are not code_name and SMILES - Added more RDKit descriptors in QDESCP with Descriptors.CalcMolDescriptors() (only for rdkit>=2023) Version 1.5.0 [`url `__] - A second ConstrainedEmbed() function with a core with no Hs was added to avoid RDKit embedding issues that show up in tricky cases - Fixed the --charge and --mult options when using xyz/sdf/pdf/gjf/com files as inputs in CSEARCH (xyz fixed in QPREP as well) - Options low_check and e_threshold_qprep were added to QPREP (create inputs only for n conformers of below a certain energy threshold) - Option nodup_check was added to QCORR (turns off the duplicate filter) - DBSTEP buried volume added in QDESCP with the qdescp_atoms option - Atomic properties of a single atom type were added in QDESCP with the qdescp_atoms option - Creation of databases for AQME-ROBERT workflows with the --robert option (True by default) Version 1.4.7 [`url `__] - QCORR is compatible with directories that contain a "." symbol - QCORR is compatible with ORCA calcs (it doesn't detect spin contamination yet) - QCORR includes hessian calculations for calcs with extra imaginary frequencies by default (new option to control this: im_freq_input) - QCORR tries to fix SCF convergence issues in ORCA by adding the SlowConv keyword - qm_end option is added after the genecp section in QPREP - Fixed a bug when using the destination option in CMIN-xTB Version 1.4.6 [`url `__] - The files and input options are compatible with partial PATHs, full PATHs, and direct names from command lines and Jupyter Notebooks - The SUMM option was fixed in CSEARCH - The files and input options now tolerate PATHs that contain directories with "." characters Version 1.4.5 [`url `__] - Suffix/prefix options work in CSEARCH, CMIN and QPREP - Automatic recognition of metals with the auto_metal_atom option - In QPREP, if qm_input starts with "p ", the Gaussian inputs starts with "#p" - CSEARCH-CREST updates the CREST outfile as the program calculates (not at the end only) Version 1.4.4 [`url `__] - When using a CSV as input, the user can specify charge and mult for each species by using the charge/mult columns - QCORR now detects duplicates including the successful calculations from previous runs - Fixed an error in full_check from QCORR when using genecp - Admits lists in command lines specified as ["X"], "[X]" and '["X"]' Version 1.4.3 [`url `__] - Return metal into RDKit mol object when using the metal_atoms option with CSEARCH-CREST - Doubles bonds do not add extra charges in metal complexes when using the automated charge calculation from SMILES - Deprotonated SiR3 groups add -1 charge to metal complexes when using the automated charge calculation from SMILES Version 1.4.2 [`url `__] - Fixed an error that raised when using CSEARCH-CREST with organic molecules - Adding more information printed when running CSEARCH - Updated README with citations from external programs - Fixed a bug during filtering of xTB conformers in CMIN (using kcal/mol instead of Hartree in the filters now) - Writing CSEARCH-CREST conformers in kcal/mol instead of Hartrees - Templates are not active when using metals with different number of ligands (i.e. if complex_type='linear' and Cu2+/CuL2 are used simultaneously) - Fixed squarepyramidal templates Version 1.4.1 [`url `__] - Changed the way xTB works in CMIN. Before, it worked through xtb-python, but in this version xtb is called through the xTB external command. This change speeds up the calculations and avoids problems for people that do not have xtb-python installed. - Fixed some bugs in the PATHs when using AQME through command lines - Updated information printed in QDESCP - Adding more error prints when no program or files are specified Version 1.4.0 [`url `__] - Fixed a bug in the automated charge and multiplicity detector for metal complexes - Adapted CREST workflows to work with metal templates - Refactored utils and rearrange files to meet code analyzer standards - The mol object that CREST uses as input now comes from the RDKit conformer generator (otherwise, metal templates aren't applied and stereochemistry information might be lost) Version 1.3.1 [`url `__] - Workflows were updated - Small fixes in CREST when using constraints - Readme was updated - GoodVibes added in installation requirements Version 1.3.0 [`url `__] - Publication version Version 1.2.0 [`url `__] - This version improves how AQME reads PATHs from arguments to make the program more robust Version 1.1.0 [`url `__] - Fixes pip install issue coming from older versions Version 1.0.0 [`url `__] - First official version of AQME ready to generate publication-quality results