|
|
|
In the picture below, certain key functions are selected of the lead identification and lead optimization process. Clinical and development phases are neglected because MDL does not provide solutions in these areas.
The pre-clinical area is important for the near future, because here cheminformatics, genomics, combinatorial chemistry and high throughput screening can make drastic improvements. Such improvements are needed to propel the industry to an seemingly impossible three to eightfold increase in output of new drugs. ( ...some companies are heading for a crash unless they can rethink their approach so completely that R&D costs and lead times plummet, generate additional sales revenues with blockbuster drugs or move into brand new markets.... there could be as few as 13 top companies by the year 2005. Sir Mark Richmond, Glaxo's former Head of Research Worldwide, goes further. "We could see just three...." taken from "Pharma 2005, An Industrial Revolution in R&D from PriceWaterhouseCoopers.)
The functions can be grouped in
Over the past 50 years, the pharmaceutical industry has concentrated on about 500 targets. The Human Genome Project opens the field to an estimate 80'000 - 100'000 new genes. The validity of this approach has to be proven first. (An interesting reading makes "Ein zweites Leben: autobiographische und andere Texte von Erwin Chargaff - Stuttgart: Klett-Cotta, 1995, ISBN 3-608-93313-1"). This is the area of genomic databases and a lot of data are available on the Internet and as well the needed analyzing tools.
However, the bottleneck seems to find the people with the knowledge and experience to exploit the incredible large data volumes that are created in this field. A company in Israel Compugen, is trying a new approach. They developed new algorithms and loan their software and people to analyze genome sequences.
See "Assay Explorer" and "Assay Development".
Companies are using two approaches, the one is running screens of any compound that becomes available, and/or using selected compounds.
The simplest selection process starts in the buying process, by elimination for instance compounds that have nitro groups. A lot is done by visual inspection. Still it cost a lot of time and effort to select only a few thousand screening compounds from one supplier database with 50'000 compounds. Clustering techniques are very helpful, and a consulting project from MDL is the Compound Selector. This is a simple application designed to cluster supplier databases. (See our offer of screening compounds!) A much more sophisticated application is MDL's Reagent Selector. This product combines access to in-house inventories and commercial databases. Chemical rules can be defined, i.e. - no metal atoms -, the hits can be clustered and the order can be generated automatically.
In summary, for effective reagent management, functions such as inventory control, reagent tracking and logistic management must adequately harmonize with synthesis planning, suppliers' offer and purchasing transactions.
Companies like Pharmacopeia and ComGenex to name only two of at least 20, provide information with their screening samples. Most of the information is calculated information. How would it be to generate very quickly and very reliable the biological activity of a sample? Impossible, not if you are satisfied with a 80-90 % success rate instead of 100%. PASS is a program developed by Poroikov V.V., Filimonov D.A. (download a paper that describes the influence of the size of the training set to the predictive power of the program, to be published.) PASS predicts about 500 activities. The program works because it uses about 30'000 descriptors, takes hydrogen atoms into consideration and has a fast hash algorithm.
Besides buying substances combinatorial chemistry is used to fill the inventory and to make specialized libraries for optimizing lead candidates. Go to the page "Combinatorial Chemistry" to see how MDL can help:
In the past biologists and chemists lead fairly separate lives, one building was the biological testing lab, the other organic chemistry. This has changed, in some companies the biologist have studied chemistry. Chemists and biologist work very closely together under one roof. It seems it was easy to bring these two groups under one roof. It was much more difficult to integrate their information systems. We will revisit this issue if we discuss the information flow in the next picture.
The next big issue is to standardize the results of a biological assay. One started with standardized templates, only to realize that this was more hindrance than progress. A lot of trial an error lead to a product like "Assay Explorer" helping the biologist to manage and to communicate their data.
With the advent of combinatorial chemistry, the development of new assays and robots came the time of dedicated HTS laboratories. These are now factories. One issue is to run plates through this factory. This means we need software that controls the robots. This software is either provided by the supplier of the robots or by specialized companies. MDL wrote the controlling software of an inventory robotic system, called Haystack, but normally, this is not the field of MDL. Hoever, MDL provides systems that enable the researcher to select the compounds, interface with the robots to load the plates and to analyze the results. All this will be incorporated into "Assay Explorer". At the time of this writing (March 2000), you would need MDL SCREEN to do the plate management.
In summary you have to solve these issues
Once a candidate seems promising you promote the substance (stuff) to focused screening. Now, the parameters of the assay are manifold and any software needs to be very flexible and you also need to interface to more analyzing tools and to statistical software packages. Data mining and visualization become more important.
Fail early, or not at all! ADME studies are extremely expensive and time consuming. Most substances are monitored by radioactive decay, this means you have to use an isotope lab, and start synthesis with the most simple substances like carbon 14 enriched CO2. Much easier is to look first to analogies in database like Metabolite and Toxicity.
You have to validate your candidate. You have to use in silico and experimental methods. Docking programs can help you to see which functional groups need modification without loosing activity. SAR studies with sophisticated software packages are possible, but are seldom done. The time it takes to make a thorough SAR study is the same as synthesizing and testing a few compounds. However, every medicinal chemist is using a SAR spreadsheet, even if he does it by hand. A SAR table is created very conveniently with ISIS/Base or ISIS for Microsoft EXCEL.
- Find new lead compounds
- Optimize lead compounds
- Find new drug families (skeletons) for a special application area
- Find new applications areas for known drug families
- Buy screening compounds
- Synthesize single compounds
- Synthesize compound libraries
- Study the metabolism of drugs
- Be aware what the competition is doing
- Study structure-activity-relationships
- Have an electronic lab journal
I am a biologist who needs to:
- Find new lead compounds
- Optimize lead compounds
- Develop assays
- Screen many compounds
- Screen few compounds
- Study structure-activity-relationships
I am a research manager who needs to:
- Produce more lead compounds
- Select the most promising projects
- Shorten the research cycle
I am an IT person who needs to:
- Buy/build information system that integrates chemists and biologist
- Develop a chemical intelligent web based information platform
- Integrate direct access to electronic journals
- Help to manage writing an electronic new drug application (NDA)
- Buy/build an inventory system
- Buy/build high throughput screening system (HTS)
- Integrate chemical ordering system with SAP
- Buy/build registration system for chemistry
- Buy/build registration system for biology
- Buy/build global search system
- Search for structures in ORACLE
- Standardize on chemical drawing tool
- Understand the underlying architecture of the MDL systems
- Use an open system architecture, muititiered, CORBA standard, platform independent, Windows based, thin clients, thick clients, call a system...
I am an assistant who needs to:
- Write papers with chemical structures
- Publish papers with structures on the Intra/Internet
- Make literature searches in chemical databases
|