Research The Bayer Scientific Magazin, Edition 28

however. “Before we synthesize drug 

candidates, we use computers to, for in-

stance, decide which of the astronomical 

number of potential compounds would be 

the most promising ones,” explains Pro-

fessor Alexander Hillisch, who leads the 

Computational Chemistry Department 

at Bayer HealthCare in Wuppertal. “For 

example, the charge of the virtual mol-

ecules plays an important role and that 

we can calculate.” The charge influenc-

es how a substance dissolves, how well 

it penetrates membranes and whether 

it could cause side effects. To be able 

to predict the charge state, Hillisch and 

his colleagues examine the functional 

groups of the molecule closely. These 

components determine the characteristic 

chemical properties of the compound. 

“The crucial factor that we use to deter-

mine the charge status is what is termed 

the pKa value of these groups,” explains 

Dr. Mario Lobell, a chemist and software 

developer in Computational Chemistry at 

Bayer HealthCare.

The pKa value of a compound can al-

so be determined experimentally, but for 

the millions of potential drug candidates  

and especially for virtual molecules which 

have not yet been synthesized, that is 

not possible. “Using our calculations, we 

can to a certain extent establish which 

molecules are promising right on the 

computer. We can then synthesize them 

in the laboratory and carry on with our 

work,” explains Hillisch. Bayer’s scientists 

use special software to calculate the pKa 

values, but at first it caused problems. 

“The programs were relatively slow, they 

weren’t very user-friendly and they were 

also relatively imprecise,” says Lobell. The 

reason was a dilemma faced by the soft-

ware manufacturers, who can only devel-

op and train their programs using publicly 

accessible molecular compounds and pKa 

data. “But these substances are very dif-

ferent in terms of their structure to our 

typical pharmaceutical compounds,” says 

Lobell. 

As a result, the predictions for medicinal 

drug candidates were automatically im-

precise. “It was as if you had crammed 

French vocabulary and then had to write 

an exam in German,” explains the Bayer 

scientist. 

To improve the data situation,  Bayer’s 

experts cooperated with the Califor-

nia-based software company Simula-

Mos doles molupturio denihil-

lore comnis sit omni

The charge of a molecule 

 determines its action

1

hour

is enough time for the new 

software to process more than 

100,000 compounds.

Source: Bayer HealthCare

Virtual chemistry for real-life treatments: Dr. Mario Lobell and Professor Alexander Hillisch (photo left, left to right) have co-developed software that 

supports optimization of new drug products, such as drug candidates for the treatment of pulmonary hypertension (photo right).

40

Bayer research 28    July 2015

---

tions Plus Inc. Together they developed 

a complex program that determines the 

pKa values of molecules significantly 

more precisely and faster: it can calculate 

more than 100,000 compounds per hour 

in unprecedented quality. “That means it 

is capable of calculating Bayer Health-

Care’s entire compound library in just 

two days,” says Hillisch. “It significantly 

accelerates and simplifies drug  discovery 

and molecule design,” adds Lobell, who 

managed the liaison with Simulations 

Plus. 

The new software uses the chemical 

structural formula of a drug candidate 

and presents the information in graph-

ic form with the relevant data listed in 

tables. It also takes into account import-

ant factors that previous programs could 

not factor in, such as the interactions 

between several functional groups. For 

this, the software has to be trained with 

as many known molecules and param-

eters as possible. Simulations Plus had 

already built pKa models using data for 

some 11,000 compounds known to them 

from the open scientific literature.  Bayer’s 

scientists contributed roughly 19,500 ad-

ditional measured pKa values for some 

16,000 pharmaceutical compounds, along 

with their molecular structures. “We were  

very impressed by the quantity and qual-

ity of data that Bayer provided us with. 

And they also gave us a huge amount of 

support during the development of the 

new model,” relates Dr. Robert D. Clark, 

Director of Cheminformatics at Simula-

tions Plus. Together, the experts in Ger-

many and California refined the model 

and closely examined any unusual values, 

thus making sure that only high-quality 

data were used. “Neither Simulations Plus 

nor we could have solved these problems 

alone,” says Lobell. 

Thanks to the innovative predictions 

methodology of the software company 

in combination with the comprehensive 

experimental database and know-how of 

Bayer HealthCare, a fundamental scientif-

ic problem has been solved. “The software 

is now commercially available worldwide. 

In this way we’re fostering progress in 

drug discovery and not just keeping this 

approach for ourselves,” says Hillisch. And 

his team is already looking at new objec-

tives, such as predicting the lipophilicity 

or water solubility of drug-like molecules 

with a high level of accuracy. 

Using mathematical models and vir-

tual patients, the researchers in Bayer’s 

laboratories have an increasingly deep 

understanding of the routes that drug 

products take through the human body 

and are thus able to optimize their drug  

candidates for patients.

What advantages does predictive software offer drug developers?

Imagine you’re standing in a completely dark room and want to 

find the exit. You’ll find it eventually just by wandering around but 

it would probably take you a long time. Wouldn’t it be great to have 

a flashlight in this situation? Like a flashlight, predictive software 

helps us to orientate ourselves when we are searching for answers 

to a whole host of questions that are vital to pharmaceutical sci-

ence. It saves money and, even more importantly, time.

What impact will new software and mathematical models have on 

medicine?

The predictions derived from these models play an important role in 

helping us understand what happens to the active ingredient of a 

pill once it has been taken by an animal or human. Computer simu-

lations of basic physiological processes enable us to reliably estimate 

dosages for initial trials in humans and to predict how effective a 

specific dose is likely to be. 

We are also seeing promising approaches for predictions of the reac-

tions in children, pregnant women and the elderly. That reduces the 

number of subjects needed to carry out appropriate clinical trials for 

these vulnerable populations.

“ Prediction software is  

like a flashlight”

Robert  

Fraczkiewicz

research

 talked to Dr. Robert Fraczkiewicz, lead scientist at the software 

developer Simulations Plus Inc., about the collaboration with Bayer.

Software trained with known 

pharmaceutical compounds

www.research.bayer.com/

virtual-liver

More information on this subject

Computer models

  DOSSIER

Bayer research 28    July 2015

41

---