Dr Julian Northen of OnyxIpca looks at the application of emerging technologies in the small molecule field
The evolution of integrated small molecule drug development
Dr Julian Northen of OnyxIpca looks at the application of emerging technologies in the small molecule field
Historically about 90 % of all marketed drugs are represented by small molecules. 1 This landscape and market is now evolving to include a significant rise in the application of biologics, such as antibody, cellular and gene-based therapies but small molecules continue to play a significant role in the production of novel treatments for disease, an example being antibody-drug conjugates( ADCs).
The evolution is not limited to the expanding range of therapies, but also to the process of development itself. Drug development is both expensive and time-consuming, in the range of $ 1-2 billion and ten to 15 years to bring a novel treatment to market. Significant effort has been focused on all areas of drug development to improve on this.
There is currently significant press coverage of the rise of artificial intelligence( AI) and machine learning( ML) to improve our ability to innovate and develop. This is increasingly true within the pharmaceutical industry and in relation to small molecule development.
In silico screening tools and AI are being combined to generate new leads in the small molecule field. These ML algorithms aim to significantly reduce the risk of failure and streamline the development process to provide structures of stable, druggable target molecules that are sensible from both a synthetic and a toxicological perspective. 2
In the early 1980s and 1990s, ligand docking and molecular modelling
"' b
....
Dose / solubility ratio
X |
Using predicted jejuna! |
t) |
solubility k. g. FaSSIF) |
|
|
Q) |
10 |
E
u
V, C:
"'
E
:::,
:x:.!:
Q) Cl.
-0 Q) ti
Q) ci:
Figure 1 – Solubility vs. permeability relationships for Classes I-IV
0.1
Good solubility and p2rmeability
Ill
Good solubility poor permeability
250 500 l, OOO
___,,..
___,,..
Figure 2- DCS put forth by Butler and Dressman
advanced rapidly for the screening of compound libraries for hits and subsequently in the optimisation of lead molecules guided by crystal structure models. This very effective tool continues to evolve. Almost every area of development has since been influenced by some form of ML and AI involvement, from route design to toxicological predictors. 3
Although this is an exciting advancement in the complex process of developing new medicines, the need for the integration of more traditional expertise remains. Route design and optimisation in particular
Ila
5,000 10,000
Solubility limited 11 b
IV
Poor permeability, poor solubility
II
100,000
Good permeability, poor solubility
is an area where a truly integrated approach is of benefit. Not all routes predicted are viable, however, and experienced chemists still play a role in selecting those more fitting to the target molecule.
A significant issue is data quality that feeds the models. The literature is rich with data of what worked, but often deficient in reporting failed or partial success in a manner that can better train ML algorithms. This area will receive significant attention in the coming years and will have an impact upon the success rate of computeraided route design.
30 SPECIALITY CHEMICALS MAGAZINE ESTABLISHED 1981