Michael A. Kappler
Daylight, CIS
The current release is Daylight Software v4.83. It is minor release that upgrades parts of the software and documentation. Supported platforms are Sun, Red Hat, SGI, and Macintosh. Areas of new development include depiction aesthetics, DayCart optimizations, the Perl wrapper, a Windows distribution, and bug fixes.
Table 1: Product Platform Matrix | |||||||||
with
|
Toolkits |
ThorServer |
MerlinServer |
Applications | |||||
32-bit | 64-bit | 32-bit | 64-bit | 32-bit | 34-bit1 | 32-bit | 64-bit | 32-bit | |
Solaris 7, 8, 9 | X | X | X | X | X | X | X | X | X |
Linux 7.1, 7.2, 7.3, E2,1, E3.0 | X | X | X | X | X | ||||
Irix 6.4, 6.5 | X | X | X | X | X | X | X2 | ||
OSX 10.2, 10.3 | X | X | X | X3 | |||||
Windows 2000, XP | X | X4 |
Partial Distribution. The partial distribution requires installation of v4.81 or v4.82, and comes with an installer that provides an interactive, simple and reversible way to migrate between versions. New in v4.83, there is an installer option that allows you to operate on either a copy of your installation, or directly on it. By default, the installer will operate on a copy and the result will be a v4.83 installation that coexists with a preexisting installation. The partial distribution form is useful for carrying forward a customized installation.
Complete Distribution. The complete distribution has no prerequisites, and is essentially v4.81 with the v4.83 patch applied to it. The complete distribution form is useful for installation from scratch and comes with Thor demo databases.
Migration Issues. Language specifications have not changed between v4.7x and v4.8x. Thor databases made with v4.7x are 100% compatible with v4.8x, so there is no Thor database upgrade procedure. A v4.7x to v4.81 upgrade procedure is required for DayCart package specifications and "backing" index object formats. It is not necessary to modify data in base tables of DayCart as part of the upgrade.
DEPICT Toolkit. Advances aimed at improving visual recognition of structure from a Daylight depiction have been developed. In v4.83, isomeric SMILES was been improved by selectively changing wedge and hash bond styles to straight lines. Also, chiral hydrogens that are not necessary for illustrating stereochemistry can be suppressed. In v4.82, control of layout with 2D-coordinates was introduced. In v4.81, display of Kekule-form became available. Collectively, these aesthetic developments produce more conventional depictions and improve structural recognition.
Table 1: Old vs. New Depictions | |||
Name | Old | New | Comments |
Demecolci | eXchange SMILES | ||
Epitestos & Estrad16a | 2D-coordinates | ||
Carbidopa | bond styles | ||
Amygdalin | hidden chiral H's |
Table 2: Historical View of Depiction Improvements | |||||
Name | v4.81 (old style) | v4.82 (2D-coordinates) | v4.83 (bond styles) | v4.83 (hidden chiral H's) | Comments |
Pencillam | better layout less bond styles hidden H | ||||
Ajmaline | better layout ring bond in plane less H's | ||||
Fosenopri | zoom out | zoom out | zoom out | zoom out | exocyclic ring bond in plane selectively less bond styles less H's |
Suloctidi | zoom out | zoom out | zoom out | zoom out | bonds between chiral centers remain in plane |
Note for Programmers: The depiction object obeys three new named properties: "old_style", "hide_chi_h", and "hlen_pct", in addition the "xsmiles" property. These are documented in the manual page for the depiction object. Layout of structure with fully or partially specified coordinates is possible. You can set all coordinates of a structure or set a substructure then use the DEPICT Toolkit to layout the remaining portion. For more information, see the manual pages on dt_setcoord and dt_calcxy.
Note for Users: Three analogous options for controlling application behavior: DEPICT_OLD_STYLE, DEPICT_HIDE_CHI_H, and DEPICT_HLEN_PCT, are obeyed by our graphical applications ( XVTHOR, XVMERLIN, and XVPCMODELS) and the Printing Package ( PRADO>, TABLET, and TREETOPS).
Note for Web Users: The three new named properties: "old_style", "hide_chi_h", and "hlen_pct", along with the "xsmiles" property are obeyed by the DayHTTP-based SMI2GIF & DEPICT applications. Layout of structure can be controlled by specifying TDT-formatted 2D-coordinates within a URL. SMILES and TDT characters that are reserved by the HTTP protocol are handled by a "Contrib" hexadecimal-encoding programs as $DY_ROOT/contrib/src/applics/http/.
Note for Cheaters: It has been empirically discovered that some of the characters used in SMILES and TDTs do not have to be encoded. Except for the triple bond (#) and the dollar sign ($), many SMILES and TDTs have been known to work directly in a URL. Some browsers convert additional characters, such as angle brackets (<,>) and update the location field "on-the-fly". This "feature" may be convenient because it skips the URL-encoding step, but compatibility varies according to browser and should be used cautiously, if at all.
Suppose we want to view morphine with 2D-coordinates and try some of the new options. We could do the following:
You should see something like this:
The same 2D-coordinates can be used for a set of stereoisomers.
Table 4: Mixing Isomeric SMILES with 2D-Coordinates
2D<-1.91,0.21,-0.60,-0.36,-0.58,-1.80,0.62,0.37,0.62,1.86,1.84,-0.28;> | |||
Name | C[C@H](N)C(=O)O | C[C@@H](N)C(=O)O | Comments |
L- & D-Alanine | no longer a problem |
DayCart Optimizations. A new, experimental optimizer package is included with DayCart in this version. It is specific to Oracle 9i and 9iR2 and gives the user the ability to query and scale the statistics provided to Oracle. Entry points have been added to the DayCart shared object file ($DY_ROOT/lib/ddlib.so), and the current "ddoptimizer" should be uninstalled before the current one is installed. This is an experimental package, and is not recommended for use with production Oracle databases. Positive testing and feedback may result in a fully supported optimizer in v4.9.
Statistics are associates with the ddpackage package and ddexact, ddgraph, ddrole, and ddblob indextypes. The user can query selectivity values and CPU and I/O costs that the optimizer provides to Oracle, which are used internally by the SQL execution engine to choose an execution plan. Selectivity values range from 0 to 100 percent. CPU and I/O costs are available for functional and index execution paths. CPU costs are reported in cycles and typically range from 100 to 10,000. I/O costs are reported in blocks. For more information, see $DY_ROOT/dcischem/README.optimizer9.
DayPerl Build. DayPerl is a wrapper which provides access to the Daylight Toolkit from Perl language programs. A completely new build procedure for DayPerl eliminates many of the Perl version related problems with the previous build and is located in the $DY_ROOT/contrib/src/perl/DAYPERL_BUILD directory. This version of DayPerl (v1.4) is supported for Perl v5.6 to v5.8. You can download the DayPerl build and compiled files from http://www.daylight.com/support/contrib/dayperl.
Windows Distribution. Daylight Software is available on Microsoft Windows in native form. Yes, it's what you wanted! This is a public beta release, meaning it's "Officially Unsupported", and some of the features are not complete or fully tested to our satisfaction. All Toolkits except Program Objects, Rubicon, and XView, are included. DayCart is available. The only binary provided is "testlicense". For more information, see the Windows-specific README file included in the release. This software is distributed as a self-extracting ZIP file and is available "upon request".
Applications
THORLOAD. The error queue is cleared after every TDT, whether or not the option to show the message is on. Valance warnings were overflowing the queue, which caused the error level would be raised to DX_ERR_ERROR and TDTs to be rejected.
XVMERLIN. "Print hitlist..." communicates datatype tags names to for depiction to tablet(1) (using -PRINT_DEPICT) by checking columns (using dt_dfnorm) for DX_THOR_USMILES, DX_THOR_ASMILES, DX_THOR_USMILESANY, DX_THOR_ASMILESANY. Previously, only columns were named SMILES, Isomer, or IsomSMILES would be depicted.
Servers
THORSERVER. The default value for the caching of hash tables has been changed from "always" to "on request". Caching used to occur with any database operation, so this avoids loading overwhelming large hash tables during creation of the database (THORMAKE) and alteration of database settings (THORCHANGE). The performance consequence of this change is small (creation of our WDI '03.2 database took less than 1% more time). If you want to cache hash tables as before, you'll need to use -CACHE_WHEN ALWAYS -CACHE_LEVEL READWRITE options.
Libraries
libdt_depict. Double bonds on tetrahedral chiral centers (e.g., [P]), may no longer be drawn with a bond style (wedge or hash), which obfuscated the bond order. Bond styles have been addressed for carbon centers, and avoiding obfuscation of the double bond is by virtue of using less bond styles. Bond style will be shown on single bonds only in the future.
Table 4: Non-carbon Tetrahedral Center: C[P@](=O)(OC)O | |||
Name | pre-v4.83 (obfuscation) | v4.83 (exposed) | Comments |
(theoretical structure) | stereochemistry is correctly displayed 50% of pre-v4.83 layout were wrong |
Layout of cis/trans double bonds has been reworked. Theoretically, 50% of exocyclic alkenes were being laid out improperly, and have been corrected. The infamous "T" indicators were always correct, now the layout and indicators are in agreement.
Table 5: Exocyclic Cis/Trans Double Orientation: C/C(=C/1\CCOC(=O)S1)/N(Cc2cnc(C)nc2N)C=O | |||
Name | pre-v4.83 (wrong) | v4.83 (right) | Comments |
Cycotiami | double bonds obfuscation may be avoided |
Fused macrocycles continue to be laid out incorrectly 50% of the time.
Table 6: Problematic Cis/Trans Double Orientation | |||||
Name | N2CCCCCCC(CC1)CC2{C}nO/C1=C/O, n=1,2,3,4 | Comments | |||
(theoretical structure) | still a problem |
Acyclic double bonds bridging two fused ring systems is a pathological problem.
Table 7: Pathological Cis/Trans Double Orientation | ||
Name | C1C2O1\C2=C3\O4C3C4 | Comments |
(theoretical structure) | still a problem |
libdw_xview. The maximum number of entries in the XView file widget has been increased from 1000 to 10K and error checking has been added.
Toolkit Routines
dt_cansmiles. The correct canonical SMILES for a molecule after resizing a ring using the toolkit ( dt_mod_on, dt_dealloc, dt_addbond, and dt_mod_off) has been corrected. Deleting a ring bond and creating a new maximum ring size didn't clear an internal state properly and the canonical SMILES was sometimes wrong.
Miscellaneous Items
Contrib. The DayHTTP CGI-based file upload ($DY_ROOT/contrib/src/c/http/http-upload.c) now works.
Linux Linker. Forward compatibility on Red Hat 9 and Enterprise 3 systems require third party open source code to resolve character type entry points (__ctype_b, __ctype_tolower, __ctype_toupper). These were eliminated from the GNU C library (glibc) in favor of localization-aware versions. Talk to us for more information.
MCL. The hitlist position and length is back, and the "Print table..." and Print status." commands are working better. This item was ready, but was inadvertently omitted from the release, and so is available "upon request".
$ cat print-status.mcl Select database $1. Print status. Create column of datatype "$SMI" named "SMILES". Create column of datatype DPN named "NAME". Create hitlist "hits". Put SMILES superstructures of "C12CCCC2C3CCc4ccccc4C3CC1" into "hits". Move to row 7. Print table from row 1 to row 10 containing SMILES NAME. Print status. Free database $1. $ mcl -i print-status.mcl wdi033%@::thor% =============================================================================== CURRENT CONTEXT: database: wdi033@vega:thor:thor =============================================================================== =============================================================================== hits 1 to 10 of 213 in hitlist "hits", database "wdi033@vega:thor:thor" ------------------------------------------------------------------------------- hit SMILES SDF name --- ------------------------------------ ------------------------------------ 1 CC12CC(O)CC2C3CCc4cc(O)ccc4C3CC1 16-ALPHA-ESTRADIOL 2 CC12CCC3C(CCc4cc(O)ccc34)C2CC(I)C1O 16-ALPHA-IODOESTRADIOL 3 CC12CCC3C(CCc4cc(O)ccc34)C2CCC1(O)C= 17-ALPHA-BROMOVINYLESTRADIOL 4 CC12CCC3C(CCc4cc(O)ccc34)C2CCC1(O)C= 17-ALPHA-IODOVINYLESTRADIOL 5 CC12CCC3C(CCc4c(Br)c(O)c(Br)cc34)C2C 2,4-DIBROMOESTRADIOL 6 CC12CCC3C(CCc4cc(O)c(O)cc34)C2CCC1=O 2-HYDROXYESTRONE > 7 COc1cc2C3CCC4(C)C(O)C(O)CC4C3CCc2cc1 2-METHOXYESTRIOL 8 COCc1cc2C3CCC4(C)C(O)CCC4C3CCc2cc1O 2-METHOXYMETHYLESTRADIOL 9 CC12CCC3C(CCc4c(Cl)c(O)ccc34)C2CCC1= 4-CHLOROESTRONE 10 CC12CCC3C(CC(O)c4cc(O)ccc34)C2CCC1O 6-ALPHA-HYDROXYESTRADIOL =============================================================================== =============================================================================== CURRENT CONTEXT: database: wdi033@vega:thor:thor column: NAME (SDF name) column: SMILES (SMILES) ==> hitlist: hits (pos 7 /length 213) ===============================================================================
Complete Information
Version 4.82. Since the previous MUG meeting, there have been two Daylight Software Releases, v4.83 (described here) & v4.82 (described at EuroMUG '03). For v4.82 information, see the talk titled What's New in v4.82 and the Version 4.82 Release Notes provided with the Software Distribution.