CODESSA 2.7.16 - Change History Authors: Alan Katritsky, Mati Karelson, Victor S. Lobanov, Roy Dennington and Todd A. Keith Copyright (c) 1994-1995 by University of Florida Portions Copyright (c) 1996-2009 by Semichem, Inc. All Rights Reserved Redistribution or resale, except by Semichem, Inc., is prohibited. *************************************************************** *Changes From Codessa-2.7.15 to Codessa-2.7.16 * *************************************************************** 1) In the CODESSA GUI, added support for importing exporting correlations to newly introduced Correlation xml files. 2) In the CODESSA GUI, added support for importing correlations from newly introduced Correlation xml files. 3) In the predict.exe program, added support for importing and applying correlations from newly introduced Correlation xml files. 4) Added support for (automatically) importing properties and / or descriptors from newly introduced structure-specific xml files. 5) Fix problem reading vibrational frequency data from Ampac-9 output files. *************************************************************** *Changes From Codessa-2.7.14 to Codessa-2.7.15 * *************************************************************** 1) Fixed another potential problem with the "Names Table" of large and / or heavily used CODESSA project file (a .cod file) becoming full, by recycling the space occupied by unused old temporary names. *************************************************************** *Changes From Codessa-2.7.13 to Codessa-2.7.14 * *************************************************************** 1) Fixed bug (unnecessary program abort) during "Predict Property" or "Plot Prediction" when a predicted structure does not have one or more of the correlation's descriptors defined. *************************************************************** *Changes From Codessa-2.7.12 to Codessa-2.7.13 * *************************************************************** 1) Fixed potential problem with the "Sets Table" of large and / or heavily used CODESSA project file (a .cod file) becoming full, by recycling the space occupied by unused old temporary sets. *************************************************************** *Changes From Codessa-2.7.11 to Codessa-2.7.12 * *************************************************************** 1) Added support for RM1 and PM6 Models in Ampac-9. *************************************************************** *Changes From Codessa-2.7.10 to Codessa-2.7.11 * *************************************************************** 1) Fixed potential problem with the "Names Table" of large and / or heavily used CODESSA project file (a .cod file) becoming full, by recycling the space occupied by unused old temporary names. *************************************************************** *Changes From Codessa-2.7.9 to Codessa-2.7.10 * *************************************************************** 1) Fixed crash problem with parsing Gaussian / NBO "Natural atomic orbitals occupancies" data that is incomplete. *************************************************************** *Changes From Codessa-2.7.8 to Codessa-2.7.9 * *************************************************************** 1) Fixed problems with predict.exe command-line utility *************************************************************** *Changes From Codessa-2.7.7 to Codessa-2.7.8 * *************************************************************** 1) Fixed a bug in the calculation of the descriptors "min(#HA, #HD) [charge type]" and "count of H-donors sites [charge type]". 2) Fixed problem on Windows Vista with not responding properly to "Cancel" with file open and save dialogs *************************************************************** *Changes From Codessa-2.7.6 to Codessa-2.7.7 * *************************************************************** 1) Made a few minor changes to make Codessa work on Windows Vista 2) Fixed potential problem with saving basic structure table to .cod file when structures are specified in a Codessa input file with a relative path and the current directory is changed to another directory prior to the save and after the load. *************************************************************** *Changes From Codessa-2.7.6 (Beta3) to Codessa-2.7.6 * *************************************************************** *************************************************************** *Changes From Codessa-2.7.6 (Beta2) to Codessa-2.7.6 (Beta3) * *************************************************************** 1) Fixed crashing problem with Data->Auto Load when auto-predicting properties from an existing correlation and a structure has one or more undefined descriptors. *************************************************************** *Changes From Codessa-2.7.6 (Beta) to Codessa-2.7.6 (Beta2) * *************************************************************** 1) Changed "Data->Auto Load / Predict" menu item to "Data->Auto Load" and added the OPTION of applying an existing correlation to the loaded set of structures, if any correlations exist. If a correlation is not applied to the loaded set of structures, the user is automatically asked to calculated descriptors for them (using a named descriptor set, for example). 2) Fixed problem with potentially using redundant structure set names "Structures (PSS 1)", etc. for auto-loaded structure sets. ******************************************************* *Changes From Codessa-2.7.5 to Codessa-2.7.6 (Beta) * ******************************************************* 1) Added ability to calculate descriptors for a set of molecules from any existing set of descriptors. 2) Added ability to use Correlation Descriptor Sets (CDS) anywhere in CODESSA where descriptor sets can be used. For example, when calculating descriptors, exporting descriptors, making new descriptor sets, etc. CDS's are automatically created, renamed or removed along with their parent correlations. For consistency, they cannot be edited directly, but can be easily saved as a regular set, if desired, and then edited. 3) Added ability to load (via Data->Auto Load / Predict) multiple AMPAC (*.out), Gaussian (*.log) or MDL Mol (*.mol) files into CODESSA using a multi-file selection dialog, instead of just with cumbersome Codessa input (*.inp) files or list files. This route is primarily intended to be used to automatically calculate descriptors and make predictions for a loaded batch of files / structures but can also be used to just quickly load a batch of files for other purposes. 4) Fixed and enhanced some "Constructed" descriptor features. 5) Added ability to handle MDL Mol (*.mol) files correctly everywhere that other major file types (AMPAC, Gaussian, Mopac) are handled, including CODESSA input list files and "Data->Auto Load / Predict" in the GUI and in the CODESSA utilities. 6) Improved width of some (not yet all) non-resizable CODESSA GUI dialogs on Windows to show full data fields by default. 7) Fixed potential problem with basic structure table of Codessa project files not always being updated properly when they should be. 8) Fixed possible crash problems when trying to read in correlations from a "Best Multilinear Regression" run that failed to produce any correlations. 9) Made default name for a correlation created from a "Correlate->Multilinear Regression" run unique and consistent with other correlation methods instead of just using the name of the method itself. 10) Fixed problem with current CODESSA project being closed and not saved after a user selects "File->New" and is asked whether to save current project and then selects "Cancel". "Cancel"ing a "File->New" will now correctly leave current project open. 11) Some intermediate data in the "Correlate->Heuristic" outpur window which was reported as an F statistic was really a t statistic (e.g., in the "Best 1-parameter correlations" section). This has been corrected. The final reported data in the window was and is correct. 12) Changed "ERROR: (SCF) ouput file is not supplied" and similar messages to just "WARNING"s. These messages often appear when quantum-chemical or thermodynamic descriptors are requested to be calculated from a data file which only contains structures, such as MDL Mol files. ************************************************ *Changes From Codessa-2.7.2 to Codessa-2.7.5 * ************************************************ 1) Fix problem with not being able to save a project file (.cod file) when a structure is not already present in the project and the structure cannot be read without error from the corresponding data source file (e.g., AMPAC .out file or Gaussian .log file). 2) Fixed "Include cross terms of selected descriptors" option in "Best multi-linear regression". 3) The descriptor "FHBCA Fractional HBCA (HBCA/TMSA) [%s]" was incorrectly named "FHBCA Fractional HBSA (HBSA/TMSA) [%s]". Corrected the name. 4) Previously, the following descriptors (as well as the corresponding fragment-based ones) were defined for molecules with one or more hydrogens having a charge less than -0.185 or greater than +0.185: "HDSA H-donors surface area [%s]" "HDCA H-donors charged surface area [%s]" "FHDSA Fractional HDSA (HDSA/TMSA) [%s]" "FHDCA Fractional HDCA (HDCA/TMSA) [%s]" "HBSA H-bonding surface area [%s]" "HBCA H-bonding charged surface area [%s]" "FHBSA Fractional HBSA (HBSA/TMSA) [%s]" "FHBCA Fractional HBCA (HBCA/TMSA) [%s]" The %s is a placeholder for various charge types ("ESP", "Zefirov", "Quantum-mechanical", etc.) This was incorrect since the descriptors only make sense for postively charged hydrogens. Changed such that the descriptors are now only defined for molecule with one or more hydrogens having a charge greater than +0.185. 5) Previously, the following descriptors were incorrectly defined for molecules with one or more negatively charged hydrogens: "HA dependent HDSA-1 [%s]" "HA dependent HDSA-1/TMSA [%s]" "HA dependent HDSA-2 [%s]" "HA dependent HDSA-2/TMSA [%s]" "HA dependent HDSA-2/SQRT(TMSA) [%s]" "HA dependent HDCA-1 [%s]" "HA dependent HDCA-1/TMSA [%s]" "HA dependent HDCA-2 [%s]" "HA dependent HDCA-2/TMSA [%s]" "HA dependent HDCA-2/SQRT(TMSA) [%s]" The %s is a placeholder for various charge types ("ESP", "Zefirov", "Quantum-mechanical", etc.) Fixed so that only positively charged hydrogens are considered in calculating these descriptors. 6) Windows Only: Fixed visualization of atomic contributions to certain descriptors (e.g., partial surface area) which previously were apparently and incorrectly dependent upon the view-orientation. Bug did not affect actual results and was Windows-specific. ************************************************ *Changes From Codessa-2.643 to Codessa-2.7.2 * ************************************************ 1) Updated the Linux and other Unix versions of Codessa to be consistent with the Windows version. Available platforms for Codessa-2.7.2 are: Microsoft Windows 95/98/ME/NT/2000/XP SGI, IRIX 6.5 IBM RS6000, AIX 5.2 COMPAQ / HP Alpha, TRU64 5.1B SUN SPARC, SOLARIS 8 HP PA-RISC, HPUX B.11.00 HP Itanium, HPUX B.11.22 Intel / AMD x86_32, Linux Intel / AMD x86_64, Linux Later OS versions should be OK. 2) Fixed a few problems reading g03 log files when calculating QM descriptors. Codessa can still read g98 log files. 3) On Windows, file dialogs are now modern, instead of the old style corresponding to the DOS 8.3 file format. File / directory paths used by Codessa are not restricted to old DOS 8.3 format. This includes launching Codessa with a Codessa project file as argument, either from the command line or via dragging and dropping. Use default extension of *.cod for Codessa project files, *.inp and *.lst for Codessa data input files (For Data->Load... and Data-> Auto Load), *.des for Codesssa descriptor files (for Descriptors->Export... and Descriptors->Import...), *.out for Ampac output files etc. File / directory paths are still limited to 256 characters. 4) Added utility codfileman.exe (see codfileman.readme), a utility program which can be used to convert a platform-dependent Codessa project file to a corresponding platform-independent text file, or vice-versa. This is useful for transferring Codessa projects from one platform to another (e.g., Windows to Linux or Linux to Windows). 5) Added saving of Cartesian coordinates of structures to Codessa project files. 6) On both Windows and Unix, use self-extracting installers to ease installation of Codessa and automate post-installation tasks. 7) Added ability to export properties while exporting descriptors, via a checkbox in the "Export Descriptors" dialog box. 8) Fixed problem with number of calculated descriptors reported in log window being different than number actually available (e.g., for correlations or export). Benign bug. 9) Fixed "Import Descriptors", including problem with imported descriptors not being loaded as a set. 10) Updated predict.exe to be consistent with Codessa, including wrt properly reading g03 log files. 11) Updated and cleaned up the docs directory and examples directory. 12) Added ability to plot results of predictions, when some-or-all of the prediction set of structures have the appropriate property available as reference. Available as Plot->Predictions on the main menu bar. 13) Added to the info on the title lines of prediction plots, including the name of the prediction set of structures and the R2 for the plot. This is useful. among other things, for comparison with the R2 for the training plot when the prediction set differs from the training set. 14) Added to / clarified info on popup when a correlation is double-clicked in "Correlations" dialog box. Fix double-click / popup ability on unix versions. Add more info about correlation to text window when "Info" (previously "Print" on Windows) button is clicked in "Correlations" dialog box. 15) Allow optional display of ESP atomic charges (instead of Mulliken charges) in molecule plot window. 16) Fix potential problems calculating Fragment descriptors from MOPAC results. 17) In Heuristic "Options" dialog box, clarified the text on some of the options and added "Max number of saved correlations" (previously hardwired to be 10) and "Min. one-parameter ANOVA F-test" (previously hardwired to be 1.0) as options. 18) When saving correlations based on a training set of structures which was constructed using "Make" ("Temporary set") or was "All structures", save the set as a new set automatically (e.g., "TSS 1", "TSS 2", etc.). The reason is that "All structures" is implicitly a dynamic set and also a correlation's training set should be saved for future reference. The first time this is done in a project, an info dialog box is shown stating this. 19) For clarity, change "Select correlations to save" caption to "Select correlations to save in set ..." in the "Save Correlations" dialog. 20) Changed Codessa versioning style from X.Y to X.Y.Z 21) On Windows, increase size of some dialogs to avoid obscured text items (e.g., descriptor names), since dialogs are not resizable in Windows version of CODESSA. ************************************************ *Changes From Codessa-2.642 to Codessa-2.643 * ************************************************ 1) Fixed expconvert.exe to handle Codessa export files with missing descriptor values for some structures as well as different ordering of descriptors among structures. 2) Added "Spreadsheet Format" checkbox to "Export Descriptors" dialog box. ************************************************ *Changes From Codessa-2.641 to Codessa-2.642 * ************************************************ 1) Fixed occasional problem reading codessa.key file. ************************************************ *Changes From Codessa-2.64 to Codessa-2.641 * ************************************************ 1) Fixed "Correlation of Properties". ************************************************ *Changes From Codessa-2.63 to Codessa-2.64 * ************************************************ 1) Fixed problem with using "Assume 0 for missing descriptor values" when running "Best Multi-Linear Regression". This caused an "error while reading in correlations" error in Codessa if a correlation found by bestreg.exe involved a descriptor with "missing" values because bestreg.exe assumed 0 for them but Codessa did not. ************************************************ *Changes From Codessa-2.62 to Codessa-2.63 * ************************************************ 1) Fixed problems reading eigenvector and initial molecule specification (with the "ATOM CHEMICAL" header) sections of Mopac2000 files. Fixed potential problems reading Dipole sections of Mopac files. 2) Added predict.exe, makeinp.exe and codkey.exe to Windows Codessa distribution, including ability of predict.exe to export descriptors to a specified text file from a set of Ampac, Mopac or Gaussian data files of from a Codessa binary file. See docs/auto.readme for details on predict.exe, makeinp.exe and codkey.exe. 3) Completed the "Data->Auto Load" option in the Windows Codessa GUI. This option is now consistent with the Unix version of Codessa. This option runs predict.exe with a chosen correlation on a set of data files specified in a list (file with the extension .lst) or a Codessa input file (file with the extension .inp). Output data files are automaticall generated, if possible and necessary, structures are automatically loaded, descriptors are automatically calculated and predictions are automatically done using the chosen correlation. 4) Fixed problem with the initial directory in file (open or save as) dialogs in Windows Codessa on Win98 / WinNT. Now, on all Windows platforms, the current directory (the last directory opened from or saved to) is the initial directory for the next file dialog. 5) Fixed bug in reading Codessa input files with lines greater than 256 in length. The line length limit is now 1024, as documented in the Codessa manual. ************************************************ *Changes From Codessa-2.61 to Codessa-2.62 * ************************************************ 1) (Unix only) Modified 'predict' and created csh scripts mkinp_m.csh and GenCodEntry_m.csh to handle automatic descriptor loading and property prediction based on a set of MOPAC output files. Unlike Ampac and Gaussian, however, the MOPAC output files will not be generated automatically if they don't exist. 2) (Unix only) Modified 'predict' to be able to simply "export" (write to a file) all possible descriptors for a set of data files - can be used for Ampac, Gaussian or MOPAC data files. 3) Fixed potential problems reading eigenvector sections of MOPAC2000 output files. 4) Added the utility expconvert.exe, a simple Windows / DOS utility program for converting an exported Codessa descriptor file ("Codessa export format") to a file in "spreadsheet format", which can then be imported into programs such as Microsoft's Excel. ************************************************ *Changes From Codessa-2.6 to Codessa-2.61 * ************************************************ 1) Fixed problem reading large Gaussian ROHF Opt + Freq .log files. 2) Fixed problem reading Gaussian polarizability and hyperpolarizability tensor elements in the correct order. *********************************************** *Changes From Codessa-2.5 to Codessa-2.6 * *********************************************** 1) Addition of Fragment, Pair and Atom descriptor capabilities. In addition to the many molecular descriptors available in previous versions of Codessa, Codessa-2.6 now allows the user to visually select a fragment, pair or atom in a molecule and calculate descriptors (e.g., fragment net charge, pair dipole, atom valency) for that fragment, pair or atom. This may be useful when a set of molecules of interest possesses a fragment, pair or atom whose descriptors correlate well with a physical property of interest. a) Fragments. A fragment can be any set of one or more atoms in a molecule. For both Ampac and Gaussian based data, there are approximately 250 fragment descriptors which can be calculated, most of which are defined similarly to the molecular descriptors. To calculate the descriptors for a fragment of a loaded molecule, select "Fragment" ("Calculate->Fragment" for Windows version) from the "Descriptors" menu. A list of available molecules pops up. Select a molecule and click "Okay". A window displaying the selected molecule will pop up. The "Options" menu can be used to toggle the display the of Hydrogen atoms, atom numbers and symbols. For Unix version, they are all displayed by default. For Windows version, they are not. To define a fragment, simply point to and click each atom of the fragment of interest (fragment atoms become highlighted). To deselect an atom, simply point to and click it again (the atom is no longer highlighted). To redefine a fragment from scratch with Unix version, select the "Define Fragment..." from the "Options" menu. To accept a fragment and proceed to the calculation of its descriptors, select "Accept Fragment..." (click "Fragment" buton for Windows version) from the "Options" menu. Selecting "Close" (clicking the "Cancel" button for Windows version) from the "Options" menu cancels the fragment selection and closes the display window. Once a fragment has been defined and accepted, a list of available descriptors to calculate is presented just as with molecular descriptors. The Prefix input box is especially important for fragment, pair and atom descriptors. By default, fragment descriptors are given the prefix "f-", but this can be changed to any string for clarity and/or to distinguish between different types of fragments. b) Pairs. The selection of a pair of atoms in a molecule and calculation of pair descriptors is similar to fragments except, of course, that two atoms must be selected and the descriptors calculated are specific to pairs (e.g., bond dipole). The default prefix for a pair of atoms with atomic symbols AB and CD is (AB-CD)- (e.g., (C-H)-). c) Atoms. The selection of an atom in a molecule and calculation of atom descriptors is similar to fragments and pairs except, of course, that only one atom is selected and the descriptors calculated are specific to single atoms (e.g., atomic valency). The default prefix for an atom with atomic symbol AB is (AB)- (e.g., (Cl)-). 2) Ability to read all Ampac data ("SCF" and "Thermo") from a single file. A single "Opt+Freq" file containing all necessary Codessa descriptors can now be generated with Ampac-6.5 and later. For Codessa input files, the column heading for such single Ampac files containing all data can be either as "AMPSCFPATHS" or "AMPTHERMOPATHS". 3) Addition of the keywords CODESSA, CODSCF and CODTHERMO to Ampac-6.5 and later. Ampac-6.5 and later now recognizes these keywords and calculates and prints all of the data that Codessa can use for "SCF" and/or "THERMO" calculations. Basically, these keywords are defined as follows: CODESSA = GRAD VECTORS BONDS PI ENPART KPOLAR NOVIS FORCE THERMO CODSCF = GRAD VECTORS BONDS PI ENPART KPOLAR NOVIS CODTHERMO = FORCE THERMO 4) Unix only By default, the Codessa utilities MakeInp.csh (mkinp) and GenCodEntry.csh (gencodentry) - used for automatic generation of ampac data files and corresponding codessa input files - use the new single ampac data file format and the CODESSA keyword. To use the old "two file" format, specify the option -o on the command line (e.g., mkinp -o aa_ampac aa_ampac.inp). Note: the "Auto Load..." feature in Codessa works by simply running MakeInp.csh for a user-specified list of Ampac input files and/or directories containing Ampac input files (mkinp_g is used in a similar manner for generating Codessa data from a list of Gaussian files and/or directories). With Ampac files and directories, "Auto Load" will only work in single file / CODESSA-keyword mode and hence will only function correctly in conjunction with Ampac-6.5 or later (unless the MakeInp.csh and GenCodEntry.csh files are modified). "The Auto Load..." feature will function correctly in conjunction with Gaussian 94 or 98. 5) Corrected reading of gaussian polarizability and hyperpolarizability tensors. ************************************************ *Major Changes From Codessa-2.06 to Codessa-2.5* ************************************************ 1) Added ability to handle Gaussian data files. See docs/gaussian.readme. Among the new features added to Codessa-2.5 is the ability to obtain molecular structures and corresponding Codessa descriptors from Gaussian (development and testing was based on g94 and g98) output (.log) files. The procedure for using Gaussian-based structures and descriptors in Codessa is similar to the AMPAC-based procedure (see the Codessa-2.06 manual and the AUTO_README file for detailed information on the use of Codessa). There are some differences between AMPAC-based and Gaussian- based Codessa, however, and these are outlined below. 1) To obtain molecular structures and most of the necessary quantum- chemical descriptor information for Codessa, the keywords "Pop=(Regular, NBORead)" and "Density=Current" should be used in the route section of the Gaussian input (.com or .gjf) file and one should specify #P to start the route section and one should add the NBO control line "$NBO BNDIDX $END" after the blank line following the molecule specification section of the .com or .gjf file. If vibrational frequencies, thermodynamic, polarizability and related descriptors are desired then one should specify the keyword "Freq" in the route section. Here are some samples: *********************************************************** %chk=h2o.chk #P HF/6-31G* Opt Freq Pop=(Regular,NBORead) Density=Current H2O 0 1 O H 1 OH H 1 OH 2 HOH OH 1.0 HOH 105.0 $NBO BNDIDX $END *********************************************************** ... #P MP2/6-31G* Opt Freq Pop=(Regular,NBORead) Density=Current ... $NBO BNDIDX $END ... #P B3LYP/6-31G* Opt Freq Pop=(Regular,NBORead) Density=Current ... $NBO BNDIDX $END Note that the use of "Density=Current" is necessary for post-SCF methods such as MP2, so that that post-SCF density is used for the NAO/NBO analysis instead of just the default SCF density. Most other Gaussian keywords may also be specified in the route section of a .com or .gjf file without affecting the use of Codessa with the corresponding .log file. There are some exceptions, however. See 4) below for automatic modification of Gaussian input files for use with Codessa. 2) Instead of using the two column-header keywords AMPSCFPATHS and AMPTHERMOPATHS in a Codessa input file, one may use the single keyword GAUSSSCFPATHS (or the single keyword GAUSSTHERMOPATHS) to specify the Gaussian .log file containing all of the descriptor data (including "quantum-chemical" and "thermodynamic") for each particular structure. For example, STRNAME PROPVALUES:bp GAUSSSCFPATHS "3-chloropropene" 318.11 gauss/bp004.log "1,3-butadiene" 268.74 gauss/bp025.log "isoprene" 307.21 gauss/bp060.log "benzene" 353.24 gauss/bp089.log "cyclohexane" 353.87 gauss/bp098.log "n-hexane" 341.88 gauss/bp119.log "m-xylene" 412.27 gauss/bp145.log "3,3-diethylpentane" 419.34 gauss/bp180.log It is not necessary to specify separate "SCF" and "THERMO" Gaussian .log files to have Codessa read in all of the possible descriptors, if all of the descriptor information for a particular structure is in one .log file. If, however, the "thermodynamic" data is in a separate .log file (one generated without Pop=(Regular,NBO)) then one may specify this with the column-header keyword GAUSSTHERMOPATHS. For example, STRNAME PROPVALUES:bp GAUSSSCFPATHS GAUSSTHERMOPATHS "3-chloropropene" 318.11 gauss/bp004.log gauss/bp004_t.log "1,3-butadiene" 268.74 gauss/bp025.log gauss/bp025_t.log "isoprene" 307.21 gauss/bp060.log gauss/bp060_t.log "benzene" 353.24 gauss/bp089.log gauss/bp089_t.log "cyclohexane" 353.87 gauss/bp098.log gauss/bp098_t.log "n-hexane" 341.88 gauss/bp119.log gauss/bp119_t.log "m-xylene" 412.27 gauss/bp145.log gauss/bp145_t.log "3,3-diethylpentane" 419.34 gauss/bp180.log gauss/bp180_t.log Note that the keywords GAUSSSCFPATHS and GAUSSTHERMOPATHS do NOT imply that one is restricted to SCF methods only in using Gaussian-based Codessa. 3) There are currently a few quantum-chemical descriptors which are obtainable from AMPAC output but not from Gaussian output. These are the Heat of Formation, the GAMMA Hyperpolarizability and the some of the "energy-partitioning descriptors", such as the "Max resonance energy" or "Min e-e repulsion" for a given atom type. The Beta Hyperpolarizability is NOT available for DFT methods and post_SCF methods by default when running a "Freq" calculation. In addition, some of the Gaussian-based descriptors are defined differently than the corresponding AMPAC-based ones. For example, the bond-orders and atomic valencies are defined in terms of the Wiberg bond-index matrix using Natural Atomic Orbitals. 4) Unix only The AMPAC automation utilities "mkinp", "GenCodEntry" and "predict" described in AUTO_README also apply to Gaussian-based Codessa with a few modifications. The Gaussian-based analog of the "mkinp" command is simply "mkinp_g", which modifies a set of Gaussian .com or .gjf files so that the corresponding .log files are fully useable by Codessa. A codessa input file with a STRNAME column and a GAUSSSCFPATHS column is created based on the found .com files. The command mkinp_g contains the additional option, -n, which specifies that the .com files and codessa input file should be prepared but the Gaussian calculations not submitted - in case further editing of the .com files and/or better job control is desired. If the -n option is not specified, then for each .com file found and modified, the corresponding Gaussian job will be submitted unless a corresponding .log file already exists. The analog of the AMPAC-based command "GenCodEntry" is "GenCodEntry_g", which functions like "mkinp_g" but acts on a single file and also has the -n option. To run the "predict" command-line program using Gaussian-based data, simply specify the option -g. See the files AUTO_README, MakeInp_g.csh and GenCodEntry_g.csh for more information on these utilities.