"The Plan"

Plans for Daylight Software in 1996 and beyond.
Dave Weininger, MUG '96

Software releases

• 4.42 -- Officially, the 4.42 Software Release is an update to 4.41. All 4.42 programs, toolkits, servers, databases and languages are are functionally identical to that v4.41 except for documented bug fixes and enhancements and all existing 4.41 licenses are valid for 4.42. You can find an informal description of this release in my "4.42 cover letter", a complete description in the Daylight 4.42 Release Notes and documentation complete in every detail and complete detailed documentation in the Daylight On-line Documentation for this release.

  We will be describing the 4.42 software this morning in our talks about the "current release". Since the 4.42 software was released recently, much of the material will probably be new to those of you who aren't beta testers. Jeremy will provide an overview of the 4.42 release. Then, after a coffee break, Craig will go into more detail about many recent enchancements to the database servers.

• 4.51 components in 4.42 -- The way things turned out, many 4.51 components were ready to go by the time 4.42 was released: the Thor and Merlin servers, the Daylight Common Gateway Interface, the HTML documentation and the HTML/FTP distribution mechanism. The 4.5x components are so much better than those in 4.41 that we just couldn't resist shipping them. Since the servers are 100% backwards compatible and everything else is new, compatibility considerations don't prevent us from doing so.

• 4.51 -- Version 4.5 will be a "minor revision" of the Daylight software. A revison allows us to change languages and fix more fundamental bugs than is possible in a fully-compatible update. Along with a large number of evolutionary changes, Version 4.5 includes two major systematic changes: a shift to HTML-oriented interfaces and the introduction of the Daylight reaction handling system.

  With 4.5, Daylight is jumping into HTML-oriented interfaces with both feet. (But not quite head-first: we'll continue to support X-Windows and Remote Toolkit interfaces.) Our commitment to HTML interfaces goes far beyond a few snappy Web pages. Virtually everything we do that involves a human-computer interface will be done via HTML. We expect to use HTML interfaces for all our documentation, programming references, both special-purpose and general-purpose database interfaces, database management, marketing, software distribution, license management, meeting registration, bug reporting, and contributed code distribution (to name a few). I will be discussing the HTML-based interfaces in detail tomorrow morning.

  Reaction handling will be completely integrated into the Daylight system, e.g., SMILES (extended to describe reactions), SMARTS (reaction patterns and transformations), Thor (reaction datatrees), Merlin (reaction superstructures, substructure, similarity searching) depiction, clustering, etc. Jack will be discussing reaction handling in detail tomorrow morning.

  Much of the 4.5x functionality is avaliable on the demo machines here at this meeting.

  Release date for 4.51 is expected mid-1996.

• 4.52 -- We expect to provide a (minor!) 4.52 update in late 1996. Release 4.52 is not expected to contain any major component changes. We're planning incremental improvements in server performance and features, additional DCGI toolkit functionality, introduction of new databases (and corresponding interfaces), support of one or more additional platforms, and the inevitable bug fixes.

  With 4.5, Daylight is jumping into HTML-oriented interfaces with both feet. (But not quite head-first: we'll continue to support X-Windows and Remote

• 4.6x -- We'll be ready for another minor software revision about a year from now.

  Assuming the DCGI interfaces in 4.5x are well-accepted, the general plan is to develop a 4.6 release which is completely server-based, i.e., interfaces will be exclusively HTML-oriented. It seems likely that the 4.6x DCGI interfaces will make use of CEX and Java.

  It may include versions operating under native Macintosh/OS and Windows/NT. If so, it will be as "first class" environments, i.e., all functionality will be supported: servers, toolkits, DCGI interfaces, etc.

  Version 4.6 may also include a revamped combinatorics package which will be integrated into the mainstream data system (more like the approach used with reactions than the current monomer-based combinatorics package).

  If things fall into place (i.e., if G*d and Anthony smile on us and Mother Nature is willing) we may be ready to release Ipcress, a 3-D shape similarity search which works on small molecules (shapes) and proteins (holes). Among other advantages, it appears that Ipcress is 3-5 orders of magnitude faster than similar existing methods.

User interfaces

We've built an HTML-oriented system for user interfaces called the Daylight Common Gateway Interface ("DCGI"). We've included the basics of DCGI in the current (4.42) software. The first real introduction of the interface will be in the 4.51 software, which will also contain conventional user interfaces (i.e., X-windows). We hope that our users will respond with unadulterated approval. If so, we plan to move ahead with it and in the best scenario, the interfaces in the 4.6 release will exclusively HTML-oriented. If not, I guess we'll have to think of something else.

We believe we have overcome all major technical obstacles to making an HTML interface a reality. There are a number of significant challenges remaining before this interface can be considered fully functional. What remains to be seen is whether such an interface will live up to its apparent promise, will become an enabling technology for chemists and will become the chemical information interface of chioce.

This interface is available here, today, on all the computers in the breakout room. We'll be talking more about it tomorrow morning in detail. Please give it a try this week and let us know what you think. We need your feedback as we never have before.

Documentation

In the current release, we also provide documentation in the previous formats, e.g., hard copy, Postscript, man(1) sources, etc. Should we continue to do so in future releases?

Databases

• ACD -- We now offer MDL's Available Chemicals Directory for use with existing Thor/Merlin systems as well as a slick HTML interface. It's usually updated twice a year. MDL imposes very restrictive rules limiting demonstrations of this database, so if you're interested, try out it out here at MUG '96.

• Aldrich -- We have the Aldrich catalog data in house, complete with structures, product descriptions and prices, directly from Aldrich. This apparently contains structures from Aldrich Flavors & Fragrances, Aldrich Catalog Handbook of Fine Chemicals, The 1995 Sigma-Aldrich Steroid Reference Collection, Sigma Aldrich Library of Rare Chemicals Structure Index and Sigma Biochemicals/Organic Compounds for Research and Diagnostic Reagents. We expect to produce an aldrich96 database to be delivered with with our 4.51 software release. The redistribution agreement and price have not been finalized but we expect it to be very reasonable.

• Beilstein -- Oh, man, we were so close to a deal allowing us to deliver the whole Beilstein database in Thor format. However, when the Beilstein Institute was bought out, negotiations went back to square one. We hope to be able to provide this database within a year, but at this point we can't promise anything.

• Infochem -- The "Infochem" database is a comprehensive database of reactions from the chemical literature. It is related to the Spresi database: all molecules which appear as reactants or products in Infochem are in Spresi. The historical subsets "ChemSynth" (70,000 reactions) and "ChemReact" (400,000 reactions) will be available with our software release 4.51. We are planning to offer the whole database (2,000,000+ reactions) later this year.

• Maybridge -- We will continue to update and support the Maybridge Catalog.

• MedChem -- The MedChem database from BioByte Corp has been updated with literature through the end of 1995. The updated database is named medchem96. It contains new CLOGP and CMR calculations which reflect the new fragment data available in Daylight Release 4.42. There are no longer "a", "b", or "c" versions of this database. medchem96 contains CLOGP and CMR values as well as 3-D coordinates computed by Rubicon.

• NCI -- About two weeks ago, the National Cancer Institute has (finally!) released some biological activity data, which we have just received. The new data consists of carcinogenicity and AIDS screening results for about 25,000 of the 125,000 non-proprietary structures in their database. We will incorporate this into the NCI database and release it as nci96 with the 4.51 software release.

• Spresi -- The spresi95 database continues to grow in popularity and utility (MP Merlin, reaction data). We will continue to distribute the spresi95 version until we get the '96 update. It now looks like future updates will be licensed directly from Viniti rather than through Infochem and that content and coverage will not change.

• TSCA -- The EPA has released a new version of the TSCA (Toxic Substance Control Act) database although we have not yet received it (as of this writing). We are uncertain whether to continue supporting TSCA. On the plus side, TSCA contains a lot (40K) of high quality structures, "official" names, CAS numbers and it's free. On the negative side, it contains no associated data and only a handful of people have ordered it. We are looking for user feedback on this one.

• WDI -- Derwent continues to provide updates for the World Drug Index of extremely high quality. We will continue to distribute, support and update this database. The current version is 95.4, distributed as database wdi954.

Platforms

We may reduce support for HP-UX to "by request".

We are considering support of other operating systems this year. An AIX port is planned: although very few of our customers are using RS/6000's, the new PPC-based machines running AIX look like they will be great servers. If so we'll be ready. With a little help from our friends at UCSF we can provide an OSF-1 port (e.g., for DEC Alpha's); however, nobody's asked for it.

We will be experimenting with native ports for Windows NT and Macintosh 7.5. The resultant ports would provide stand-alone "PC" capability and may even offer some performance improvement over a remote-toolkit based system.

The current view is summarized in this table: