UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
Artificial intelligence in drug discovery
1. Department of Pharmaceutical Analysis
National Institute of Pharmaceutical Education and Research (NIPER)
Balanagar, Hyderabad Telangana– 500037
Prepared by: Ravindra Babu
PA/2019/118
GE-511
4. ARTIFICIAL INTELLIGENCE (ai)
Artificial intelligence is field of computer science that is
associated with the concept of machines “ thinking like
humans” to perform tasks such as learning , problem
solving , planning , reasoning, identifying patterns.
Artificial intelligence come up with two subfields
Machine learning
Deep learning
5. MACHINE
LEARNING
• It is technique of parsing data, learn from that data and then
apply what they have learned to make an informed decision.
• E.g.; Amazon product recommendations
• If a machine learning model returns an inaccurate prediction
then the programmer needs to fix the problem.
DEEP
LEARNING
• Deep learning is subset of machine learning where artificial
neural networks, algorithms inspired by human brain are used.
• Allows machines to solve complex problems and fix the
problems by it self. e.g. Automatic car driving system
ARTIFICIAL
INTELLIGENCE
• AI is ability of computer program to function like a human
brain
• Actually deep learning and machine learning both are subsets
of AI.
• They are used as a way to achieve AI.
6. Common examples of AI
Recommendation systems give you
recommendations based on your personal preference
Spam filtering
Face book brings you news based on what you and
your friends like
Google maps not only routes you efficiently to your
destination , it also gives notification about traffic jams,
road blocks etc.
7. common examples of AI
Gmail sorts email into 4 different tabbed categories, and
sends the spam mail to a separate folder.
The program helps your emails get organized so you can
find your way to important communications quicker.
8. Predictive searches are based on data that Google
collects about you, such as your location, age, and other
personal details.
Using AI, the search engine attempts to guess what you
might be trying to find
9. IDENTIFICATION OF NOVEL
CHEMICAL COMPOUNDS
HIT IDENTIFICATION
LEAD IDENTIFICATION
LEAD OPTIMISATION
HIT EXPANSION
IN VIVO ACTIVITY ASSESSMENT
11. DRUG DEVELOPMENT PROCESS
The feedback-driven drug development process starts from
existing results obtained from various sources such as high-
throughput compound and fragment screening,
computational modelling and information available in the
literature. This inductive–deductive cycle eventually leads to
optimised hit and lead compounds.
The first step in drug development is the identification of
novel chemical compounds with biological activity. This
biological activity can arise from the interaction of the
compound with a specific enzyme or with an entire
organism.
12. DRUG DEVELOPMENT PROCESS
The first compound that shows activity against a given biological
target is called a ‘hit’.
The identification of a lead molecule is the second step in
drug development. A lead is a chemical compound that shows
promising potential that can lead to the development of a new
drug as a treatment for a disease.
Once a lead compound has been found, its chemical structure is
used as a starting point for chemical modifications with the
objective of discovering compounds with maximal therapeutic
benefit
During the process of lead generation, hit molecules are
systematically modified to improve their activity and selectivity
towards specific biological targets, while reducing toxicity.
13. DRUG DEVELOPMENT PROCESS
The chemically related compounds derived from a hit are called
analogues and the process is referred to as hit expansion
To increase the synthetic throughput, chemists focus on a specific
reaction or set of reactions to assemble building blocks together
to make a series of analogues quickly. A ‘building block’ is a
compound that possesses a reactive functional group and atoms
that interact with the active site of a biological target
This active site is a specific region in the biological target to
which the compound (or substrate) binds through interaction
forces. The binding of a substrate to an active site can be
visualised as ‘lock and key’ or ‘induced-fit’ MODELS.
15. APPLICATIONS OF AI IN DRUG DEVELOPMENT
AI in
understanding the
pathway or finding
molecular targets
AI in finding the
hit or lead
AI in synthesis of
drug-like
compounds
Predicting the
mode-of-action of
compounds using
AI
AI in selection of a
population for
clinical trials
AI in drug
repurposing
AI in poly
pharmacology
16.
17. AI in understanding the pathway or finding molecular targets
This was possible owing to the incorporation of genomics
information, biochemical attributes and target tractability.
One study determined the plausibility of predicting therapeutic
targets using a computational prediction application known as
‘Open Targets’ – a platform consisting of gene–disease
association data.
It was reported that animal models exhibiting a disease-relevant
phenotype with a neural network classifier of >71% accuracy
provided the most predictive power .
18. AI in synthesis of drug-like compounds
TRADITIONAL SYSTEM
Drug-like molecules are compounds that obey Lipinski’s rule of five
The first step in the retro synthetic approach is to analyze the target
compounds recursively and to sequentially convert them into smaller
fragments or building blocks that can be easily purchased or
prepared .
The second step is to identify the reactions that will convert these
fragments into target compounds. The second step is the most
challenging because it is difficult for the human brain to interrogate
the vast number of relevant organic reactions available in the
literature to pick the most plausible reaction.
19. The voids in organic synthesis are mainly the result of unpredictable steric and
electronic effects and incomplete understanding of the reaction mechanism .
AI would aid in predicting the best sought-after reactions by filling the voids that
cause high failure in expected organic synthesis
Seglar et al. have developed a new AI platform named 3N-MCTS, which combines
three different deep neural networks with Monte Carlo Tree Search (MCTS) for
CAOCS
This platform can filter out the most promising building blocks and select only
well-known reactions for the synthesis of target compounds
The platform, 3NMCTS, was proven to be much faster and better than that of
traditional computer-assisted retro synthesis systems.
20. AI in finding the hit or lead
AI systems can reduce the attrition rates and the R&D expenditure
by decreasing the number of synthesized compounds that are
subsequently tested in either in vitro or in vivo systems
ML techniques and predictive model software also contribute to the
identification of target-specific virtual molecules and association of
the molecules with their respective target while optimizing the safety
and efficacy attributes.
DL becomes useful in instances where structural data are insufficient.
Thus, phenotypic data or disease, biology or molecule network-based
algorithms can be used.
21. AI in selection of a population for clinical trials
An ideal AI tool to assist in clinical trials should recognize the disease
in patients, identify the gene targets and predict the effect of the
molecule designed as well as the on- and off-target effects .
The development of AI approaches to identify and predict human-
relevant biomarkers of disease allows the recruitment of a specific
patient population in Phase II and III clinical trials
22. Predicting the mode-of-action of compounds using AI
AI platform that can predict the on- and off-target effects and in
vivo safety profile of compounds before they are synthesized.
few examples of such platforms are
(predicts toxicity of new compounds)
PrOCTOR (predicts the probability of toxicity in clinical trials)
The availability of such platforms reduces the drug development
time, R&D costs and attrition rates
23. AI in drug repurposing
The concept of applying an existing therapeutic to a new disease is
advantageous because the new drug is qualified go directly to Phase II
trials for a different indication without having to pass through Phase I
clinical trials and toxicology testing again .
In a study performed by Aliper et al., it was demonstrated that DNNs
could classify complex drug action mechanisms on the pathway level,
thus classifying drugs into therapeutic categories according to efficacy,
therapeutic use and toxicity.
24. AI in polypharmacology
One-disease–multiple-targets is termed poly- pharmacology
Many databases, such as ZINC, PubChem, Ligand Expo, KEGG,
ChEMBL, DrugBank, STITCH, BindingDB..etc.
AI could be used to probe these databases to design poly
pharmacological agents.
25.
26. STARTUP COMPANIES
Six startups providing an AI solution applicable during a
particular research phase of the drug discovery & development
process are highlighted as follows:
Generating novel drug candidates: Atomwise, TwoXar,
ReviveMed
Understanding disease mechanisms: Phenomics AI, Structura
Biotechnology
Aggregating and synthesizing information: Arpeggio
Biosciences
27. DRAWBACKS OF AI IN DRUG DISCOVERY
As is the case with any advance that brings a paradigm shift in our
understanding of an existing technology, AI still cannot replace a
human scientist entirely in the process of drug discovery.
AI predictions are as good as the algorithms used to investigate a dataset.
The algorithm should clearly lay out the criteria that should be used to parse
out meaningful information when the results are in the ‘gray zone’ of
interpretation.
AI can suffer from algorithm bias, where the creators’ own bias manifests
itself in the way information is processed to generate predictions. Therefore,
the process is not entirely objective.
While the cost of supercomputing and high-throughput screening has
decreased appreciably over the past decade, establishing these pipelines still
requires significant investment.
Ultimately, predictions made by a computer have to be verified by a scientist
to make sure they are valid
28.
29. ETHICAL ISSUES AND REGULATIONS
The algorithms used today have not been optimized for the
definitions of fairness, they have been optimized to do a task.
Since 2017, FDA approved AI algorithm in cardiac imaging
,however few formal regulations around AI are available- when
they exist at all applications of AI will rise many ethical questions
particularly in an event of an error in AI diagnosis and role of
each stakeholder in construction in AI devices.
There is no oversight that virtual clinical trials could replace
earlier clinical trials.
30. CONCLUSION
Currently, there are no developed drugs that have utilized AI approaches
but, based on the advances it is likely that it will take a further 2–3 years
for a drug to be developed.
“You have microwave and you have a coffee machine…and you have this
and that, but none of it actually cooks you the dinner…..you need to put it
all together to make a dinner, even though all these things can help you to
do it faster and better.”
However, for an individual to be efficient in drug development using AI, the
individual should know how to train algorithms, requiring domain
expertise. This creates the suitable workspace whereby AI and medicinal
chemists can work closely together.
31. REFERENCES
1. Segler, M.H.S. et al. Planning chemical syntheses with deep neural networks
and symbolic AI.,2018 Nature 555, 604–610
2. Kit kay mark, Mallikarjuna Rao Pichika, artificial intelligence in drug
development: present status and future prospects, Drug discovery today,2019,
24,773-781
3. Aliper, A. et al., Deep learning applications for predicting pharmacological
properties of drugs and drug repurposing using transcriptomic data. Mol.
Pharm.2019, 13, 2524–2530
4. Ferrero, E. et al. In silico prediction of novel therapeutic targets using gene-
disease association data. J. Transl. Med., 2017,15, 182
5. Alex zhavoronkov, Yan A. ivanenkov, Deep learning enables rapid identification
of potent DDR1 kinase inhibitors, Nature Biotechnology,2019,37,1038-1040