 |
 |
MATRIX concentrates on two core issues in drug development:
HOW CAN THE QUALITY AND SPEED OF THE DRUG DEVELOPMENT PROCESS BE IMPROVED?
And, more specifically,
HOW CAN THE RISK OF FAILURE OF A DRUG CANDIDATE AT ONE OF THE LATER PHASES IN THE DISCOVERY PROCESS BE SYSTEMATICALLY REDUCED?
In trying to define what precisely causes a particular disease, most research attempts to reduce the immense complexity of the various systems involved down to a small number of definable constituent parts. The aim is to come up with a relatively simple target, which is then based on the interaction of one compound with a single molecular structure.
|
Figure 1: Drug development reduces complex environments down to simple systems.
|
The models developed with this approach have the serious disadvantage of only reflecting a (too) small part of reality. Many of the aspects that are neglected later prove to be the cause for the high failure rates at the more advanced development stages.
We take complex in-vitro environments into account right from the beginning of a project. This means that we test and optimise the compounds both with regard to their activities vis à vis the target and in terms of their specificity, toxicity, ADME parameters, permeability and many more.
The method of successive filtering with more and more criteria (especially popular since the introduction of HTS) is replaced at MATRIX by the simultaneous computer modelling and optimisation of many test systems using relatively small amounts of data. The powerful algorithms we have developed for this purpose are capable of harvesting results from all the data available, including both the successes and the failures.
|
|
CADDIS stands for Computer Assisted Drug DIScovery, and is the title we have given to our proprietary drug development process.
At the beginning of a project, some boundary conditions are defined: the set of criteria the desired compounds are to fulfil (and thus the assay systems to be used) and the class of chemical compounds from which the compounds are to be selected (RNAs, SOMs, peptides,
).
|
Typically, we start with an initial set of 200 compounds, which are selected as representatives of the chemical class under consideration.
These compounds are synthesized and experimentally tested in all assay systems with regard to their actual activities - computer modelling is not used at this phase |
 |
|
Figure 2: The CADDIS process.
This table with the chemical structure of the compounds on one side and their action in real experiments on the other side are
fed into our computer systems, which proceed to generate the initial models of the structure-activity relationship. |
|
Based on these models, the scope of allowed compounds (those capable of synthesis) is computed and promising candidates as well as compounds leading to a maximum of new information are proposed for synthesis in the next cycle. Typically ~100 compounds are proposed per cycle.
Again these new compounds are synthesized and tested in all assay systems. Some aspects of the first models will turn out to be successful, whereas others will be inferior. Based on the new data, improved and refined models are generated.
This cycle of synthesis and experiment in the laboratory on one side and modelling and computer-based optimisation on the other side is typically repeated 6-8 times. The result is one (or even more) compounds which are optimised with respect to many criteria, so called Multi-Objective Lead(s) (MOLs).
To summarize, the advantages of the CADDIS process are:
|
 |
In total only ~1000 molecules are synthesized and analysed. Therefore: |
|
More complex and realistic assay systems are possible. |
| |
More complex chemical structures can be investigated. |
|
All major aspects up to the pre-clinical phase are covered right from the beginning. |
|
The process is robust and fault-tolerant. |
|
It is possible to define and therefore exclude parts of the search space from the process. This can help minimize patent risks. |
|
