drug discovery Archives - Pharmacelera | Pushing the limits of computational chemistry

Pharmacelera and eMolecules Launch Integrated Solution for Faster Hit Discovery

Enric Herrero — Mon, 13 Oct 2025 15:05:04 +0000

California, US & Barcelona, Spain — October 13^th, 2025 —

Together with eMolecules we announce the launch of a new integrated solution combining ExaScreen® with the eMolecules eXplore/Synple library. The partnership delivers researchers a streamlined path from virtual screening to physical compounds, accelerating early-stage drug discovery.

The pharmaceutical industry increasingly relies on large, diverse libraries to identify novel hits. Yet, efficiently navigating chemical space while ensuring synthetic feasibility remains a key challenge. The new solution combines Pharmacelera’s Quantum-Mechanics (QM) and Machine Learning (ML)-driven algorithms with the eXplore/Synple library’s curated, synthetically tractable compounds, enabling scientists to quickly identify, source, and test new chemical matter:

Accurate screening of tractable chemical space using ExaScreen®
Rapid access to compounds for experimental validation via eMolecules
Novel IP opportunities through exploration of untapped chemical diversity
End-to-end workflow from computational predictions to physical samples

“Our collaboration with Pharmacelera brings a powerful combination of computational precision and practical compound access,” said Jeff Desroches, SVP Corporate Development.

“Partnering with eMolecules aligns perfectly with Pharmacelera’s strategy of working with leading organizations that complement our technology and expertise,” said Rémy Hoffmann, Chief Business Development Officer at Pharmacelera.

—

About eMolecules

eMolecules is driven to improve the human condition by enabling scientists to accelerate their research to find effective therapeutics. To achieve this, eMolecules provides business intelligence data and integrated ecommerce software for screening compounds, chemical building blocks and primary antibody supply chains. These tools, combined with their acquisition, aggregation and analytical services, greatly empower drug discovery researchers working in the pharmaceutical, biotechnology, academia, CRO and agrichemical industries. eMolecules was founded in 2005 at its San Diego, California, USA headquarters and has offices and laboratory space in San Diego and London, UK, employing nearly 60 people, globally.

About Pharmacelera

Pharmacelera is a deep-tech science-first company founded by experienced drug hunters, high-performance computing engineers, and leading academic researchers. The company has developed a proprietary in-silico platform that integrates accurate 3D Quantum-Mechanics (QM) models with advanced Artificial Intelligence (AI) algorithms to design novel, diverse, and high-quality molecules from ultra-large and previously untapped chemical spaces. Pharmacelera provides access to its cutting-edge technology for HitId, H2L and LO through yearly software licenses and it also offers AI-driven drug discovery services spanning the entire small-molecule pipeline. The company has a growing customer base across Europe and the United States, including several top-tier pharmaceutical companies.

The post Pharmacelera and eMolecules Launch Integrated Solution for Faster Hit Discovery appeared first on Pharmacelera | Pushing the limits of computational chemistry.

Domainex and Pharmacelera Join Forces to Accelerate Discovery of Molecules Targeting Transmembrane Proteins

Enric Gibert — Wed, 27 Aug 2025 07:13:33 +0000

Cambridge, UK & Barcelona, Spain — August 27^th, 2025 — Domainex, a leading integrated drug discovery services company, and Pharmacelera, a pioneer in AI-based molecular modelling technologies, today announced a strategic collaboration to support discovery programmes focused on transmembrane proteins such as G protein-coupled receptors (GPCRs) and ion channels—critical therapeutic targets implicated in a broad range of diseases.

Transmembrane proteins represent over 60% of current drug targets but pose significant challenges for drug discovery due to their structural complexity and instability outside the lipid membrane environment. By combining their complementary technologies and expertise, Domainex and Pharmacelera aim to overcome these hurdles and accelerate the discovery of novel, high-quality drug candidates.

Under the partnership, Domainex will contribute its Polymer Lipid Particle (PoLiPa) technology (which stabilises membrane protein targets by encapsulating them in polymer nanodiscs, allowing purification in their native state for screening), alongside its Direct-to-Biology (D2B) platform for high-throughput synthesis and biological testing. Pharmacelera will integrate its state-of-the-art AI-driven platforms—exaScreen® and PharmScreen®—for advanced molecular modelling, virtual screening, and library design. Together, the companies will offer a comprehensive and seamless solution to identify and optimise hits against these challenging targets.

“This collaboration between Domainex and Pharmacelera combines innovative technologies and represents optimal approaches for drug discovery programmes and is part of our commitment to achieve faster and successful approvals for our customers.” said Dr Hayley French, CEO of Domainex.

“We are excited to partner with Domainex,” said Dr. Enric Gibert, CEO of Pharmacelera. “This partnership reflects our shared commitment to advancing science and bringing impactful therapeutics in challenging areas. By pairing our cutting-edge AI models with Domainex’s experimental data and technology platforms, we can accelerate the path from concept to candidate.

—

About Domainex

Domainex is a multi-award-winning, integrated drug discovery service partner which provides tailored discovery solutions from target expression through to the identification of pre-clinical candidates. Our world-leading experts accelerate research projects by combining a problem-solving approach with cutting-edge technologies such as PoLiPa and D2B. With deep expertise across a wide range of target classes and therapeutic areas, and a core focus on the hit identification and hit-to-lead stages, we deliver high-quality results that support confident, timely decision-making for our partners. Domainex operates from state of the art facilities in the Cambridge area, UK. The company has also expanded internationally and has opened an office in Cambridge, MA, to service the thriving biotechnology industry in North America. Further information about Domainex and our award-winning lead discovery services can be found at www.domainex.co.uk.

About Pharmacelera

Pharmacelera develops advanced computational tools for the discovery of novel hits using accurate Quantum-Mechanics (QM), Artificial Intelligence (AI) and High-Performance Computing (HPC). The company’s products PharmScreen®, exaScreen® and PharmQSAR use 3D molecular descriptors derived from Quantum-Mechanics (QM) calculations to mine an unexplored chemical space and to identify hits uncovered by traditional algorithms. Pharmacelera is a private company founded in 2015 and based in Barcelona, Spain. The company works with several big pharma and biotech organizations across Europe and the United States.

The post Domainex and Pharmacelera Join Forces to Accelerate Discovery of Molecules Targeting Transmembrane Proteins appeared first on Pharmacelera | Pushing the limits of computational chemistry.

Pharmacelera and UMass Chan Medical School Join Forces in Drug Discovery

Enric Gibert — Mon, 14 Jul 2025 13:13:05 +0000

We are thrilled to announce a collaboration between Pharmacelera and the UMass Chan Medical School aimed to support early-stage drug discovery projects.

Through this collaboration, Pharmacelera will use its cutting-edge computational chemistry technology, AI-driven molecular modelling platforms and industry-based expertise in medicinal chemistry to support several early-stage drug discovery programs targeting different families of targets with the aim of enhancing hit identification, lead optimization, and chemical space exploration.

“This partnership reflects our shared commitment to advancing science and bringing impactful therapeutics closer to the clinic,” said Enric Gibert, CEO of Pharmacelera. “Our technology is designed to push the frontiers of early drug discovery, and we are proud to support UMass Chan’s world-class research teams.”

“We are excited to partner with Pharmacelera to integrate their advanced modelling tools into our drug discovery workflows,” said Huseyin Mehmet, Executive Director, New Ventures, BRIDGE Innovation & Business Development of the UMass Chan Medical School. “This collaboration will enhance our ability to discover and optimize novel compounds for unmet medical needs in the areas of cancer and ALS.”

Stay tuned as we share more about our joint projects and scientific milestones in the coming months.

—

About UMass Chan Medical School

UMass Chan Medical School, one of five campuses of the University of Massachusetts system, comprises the T.H. Chan School of Medicine, the Morningside Graduate School of Biomedical Sciences, the Tan Chingfen Graduate School of Nursing, ForHealth Consulting at UMass Chan Medical School, MassBiologics, and a thriving Nobel-Prize-winning biomedical research enterprise. UMass Chan is advancing together to improve the health and wellness of our diverse communities throughout Massachusetts and across the world by leading and innovating in education, research, health care delivery and public service. It is ranked among the best medical schools in the nation for primary care education and biomedical research by U.S. News & World Report. Learn more at www.umassmed.edu.

About Pharmacelera

The post Pharmacelera and UMass Chan Medical School Join Forces in Drug Discovery appeared first on Pharmacelera | Pushing the limits of computational chemistry.

A new standardized protocol for the preparation of large 3D fully enumerated compound libraries

Enric Herrero — Thu, 07 Nov 2024 10:39:40 +0000

By Nicola Scafuri and Ana Caballero

A larger number of ligands in the virtual screening library increases the chances of identifying ligands that are more potent, selective, or possess improved physicochemical properties. Mining such chemical spaces using the 3D representation of molecules has shown to be a successful approach, however, a reliable screening is only feasible when the 3D library is properly prepared. Getting ready a 3D library from a 2D representation is not trivial, since several chemical aspects must be considered, for example:

A single 2D molecule can exist in multiple protomeric and tautomeric forms at a given pH, each with a distinct distribution.
Molecules with chiral centers can exist in various 3D stereoisomeric forms.
All potential 3D conformers must be thoroughly considered.

PharmScreen^®, our field-based virtual screening software, has exhibited a promising performance in identifying novel hits within 3D libraries with similar physic-chemical properties to reference compounds. Thanks to a unique and superior 3D representation of molecules based on electrostatic, steric, and hydrophobic interaction fields derived from semi-empirical Quantum-Mechanics (QM) calculations, the internal and external benchmarks have identified a promising number of novel and diverse hits for several targets. Notwithstanding, even these encouraging results rely on the adequacy of using a 3D accurately prepared library.
We have developed an internal protocol for preparing 3D libraries suited for ligand-based virtual screening (LBVS) and structure-based virtual screening (SBVS), such as docking campaigns, and pharmacophore modeling solutions. It aims to establish a standardized and easily reproducible protocol for preparing 3D libraries.
The protocol integrates internal scripts and the PharmScreen^® software, and it was initially utilized to prepare the Enamine Screening Collection library. This library is one of the world’s largest screening compound libraries, boasting over 4.4 million unique compounds. The protocol involved the generation of different protomers and tautomers at pH 7.4, and all possible stereoisomers and conformers. As a result, our protocol ensures an extensive, high-quality chemical space, to deliver unique and tailored drug discovery solutions. Indeed, we have now one of the largest and most up-to-date 3D screening libraries of synthesizable compounds for early drug discovery projects, featuring:

Up to 270 million conformers for LBVS
Up to 7.9 million 3D stereoisomers for docking campaigns, optimized for pharmacophore modeling solutions
In addition, a complete screening of this library can be achieved in just 25 hours using PharmScreen^®

Our protocol can also filter the prepared 3D library based on drug-like properties to focus the hit ID towards the drug-like space, enhancing efficiency in the execution of CADD projects.
This robust protocol has shown to be extensible to other commercial libraries, for example, the Molport Screening Compounds library, further expanding the chemical space from which we can extract novel hits.

Interested in the application of QM methods to drug discovery? Pharmacelera software uses a unique 3D representation of molecules based on electrostatic, steric and hydrophobic interaction fields derived from semi-empirical QM calculations. Discover PharmScreen^®, exaScreen^® and PharmQSAR^®. Need a customized solution for your drug discovery project? Contact our team to arrange a call and discuss your current challenges.

The post A new standardized protocol for the preparation of large 3D fully enumerated compound libraries appeared first on Pharmacelera | Pushing the limits of computational chemistry.

Using Artificial Intelligence for faster Quantum Mechanical parametrization

Enric Herrero — Thu, 24 Oct 2024 09:51:56 +0000

By Carlos Cruz and Ana Caballero

The success of a ligand-based virtual screening campaign relies on their molecular descriptors, in this sense, Quantum Mechanics (QM) offer higher accuracy by capturing detailed electronic properties, the influence of 3D conformation, and key interactions that classical descriptors often overlook. Pharmacelera´s property molecular descriptors exploits self-consistent reaction ﬁelds methods to define hydrophobic topologies from atomic contributions.

Our unique descriptors are calculated using accurate QM methods, through the Recife Model 1 (RM1) parametrization of the Miertus Scrocco Tomassi (MST) model, which is known to be a reference method due to its good balance between accuracy and calculation time. However, QM calculations are still time-consuming, especially for huge chemical spaces. To handle this challenge, our team has developed a Machine Learning model for predicting atomic logP, to determine if exploring huge chemical spaces in a reduced amount of time is possible.

Our model includes physical descriptors (ex. topological, steric, and electrostatic descriptors), transferring information about the 3D environment of the atom to the model. The precise description of each atom gave us the possibility to develop and validate a model able to predict 3D atomic contributions to logP values with an R^2 >0.9 for any kind of neutral drug-like molecule 2.000 times faster than the calculations using the RM1 parametrization (see graphics (A) and (B))

Using this atomic description, we have accurately extended it to predict other atomic properties. For example, graph (C) shows the excellent results obtained for Mulliken Charges derived from Density Functional Theory calculations from the QMUGs library.

Need a customized solution for your drug discovery project? Contact our team to arrange a call and discuss your current challenges.

The post Using Artificial Intelligence for faster Quantum Mechanical parametrization appeared first on Pharmacelera | Pushing the limits of computational chemistry.

Quantum mechanical-based strategies in drug discovery

Fernando Martín — Thu, 22 Aug 2024 09:53:21 +0000

By Tiziana Ginex and Fernando Martin

The ever-increasing accessible chemical space opens the door to the search for new chemical matter for drug discovery. However, this also poses a challenge for computer-aided drug design methods. Quantum mechanical (QM) methods provide a chemically accurate description of molecular properties, albeit restricted to small size systems. The availability of high-quality QM-based descriptors implemented in refined algorithms and combined with efficient computational protocols can help to prioritize hits, avoiding the occurrence of bias artifacts in chemical library screening.

Different efforts are underway to apply accurate methods to the ever-expanding accessible chemical space: (i) the development of computationally efficient semiempirical methods as well as the calibration of multiscale QM/MM methods, (ii) the redefinition of physics-based force fields tailored to QM, suitably refined to provide an accurate description of the complex network of intermolecular interactions, and (iii) the generation of QM-assisted machine learning (ML) models.

In this review article, the authors summarize relevant advances in the application of the above QM-based methods to the characterization of bioactive species, structure-guided hit-to-lead optimization, and the identification of molecular features of bioactivity.

Read the article

Need a customized solution for your drug discovery project? Contact our team to arrange a call and discuss your current challenges.

The post Quantum mechanical-based strategies in drug discovery appeared first on Pharmacelera | Pushing the limits of computational chemistry.

How can we screen 31 billion compounds? Divide and conquer!

Fernando Martín — Tue, 30 May 2023 13:28:37 +0000

Enric Herrero

May 30th, 2023

Commercial chemical libraries have witnessed remarkable growth in recent years, resulting in an unprecedented increase in size and diversity. With advancements in high-throughput synthesis and combinatorial chemistry techniques, compound providers like Enamine have expanded their collections of small organic molecules to meet the escalating demands of the pharmaceutical industry. This exponential growth has provided researchers worldwide with access to an extraordinary wealth of chemical diversity, facilitating the discovery and development of novel therapeutic agents.

However, the significant growth in size and diversity of commercial chemical libraries has rendered previous methods of virtual screening impractical, especially for accurate 3D methods. The sheer volume of compounds amassed within these libraries presents immense challenges in terms of storage and computational costs. Taking as a reference a compressed SD file containing multiple stereoisomers and conformers of a molecule of 48KB, the fully enumerated library of Enamine REAL (31 billion compounds) would require a storage capacity of 1.36 PB of data! This is almost 1400 hard drives like the one that you have in your laptop! Following the same example, if we assume that processing this single molecule requires 3.6 ms in your laptop, this will mean 3.5 years of calculations to perform a screening!

Luckily, several methods have been proposed that rely on the way these huge libraries are created, which is combining a set of building blocks to generate new molecules. Figure 1 shows an example of the value of using building blocks to perform a screening, for simplicity we will assume that each building block is a synthon and that all of them can be combined. In this example we have a building block library of 3 building blocks that can generate a chemical space of 9 molecules (combining all against all). If we apply the traditional approach (Brute force) we would compare our reference structure against each of the molecules of the library, this is 9 comparisons. However, if we perform a smart enumeration taking advantage of how this library has been created, we can reduce the computing cost. In this case what we would do is to partition the reference structure in two fragments and instead of comparing against all the enumerated library we perform the comparison against the building block library, this is 3 comparisons. Since we have two reference fragments we need to perform this operation twice, resulting in 6 comparisons, 3 less than in the brute force approach. This difference in the number of comparisons does not seem large but if we translate this example to a building block library of 100K building blocks, in the brute force approach we would need 10 million comparisons vs 200 thousand for the smart enumeration approach.

If we plot the computational cost projections of both methods for different library sizes (Figure 2) we can see how the scalability of the smart enumeration approach is much better than the brute force approach and, therefore, is much more suitable for the chemical libraries of the future.

Overall, we have seen that the rapid growth of commercial chemical libraries represents a challenge for virtual screening tools. The size and diversity of these libraries have made traditional screening methods impractical due to storage and computational costs. However, the use of smart enumeration based on building blocks offers a more efficient approach. By leveraging the way these libraries are created, researchers can significantly reduce the number of comparisons needed for screening. This smart enumeration approach shows better scalability and is considered more suitable for future chemical libraries, offering computational efficiency compared to brute force methods.

The post How can we screen 31 billion compounds? Divide and conquer! appeared first on Pharmacelera | Pushing the limits of computational chemistry.

General Inception and Pharmacelera Partner to Advance Drug Discovery with Exponential Screening Capabilities

Fernando Martín — Tue, 09 May 2023 14:27:06 +0000

3D quantum-mechanics, molecular descriptors, and artificial intelligence provide unique capabilities in the chemical space to identify and qualify novel hits

PALO ALTO, Calif and BARCELONA, Spain: General Inception (GI), a global company Igniter, and Pharmacelera, a Barcelona-based deep-tech company, announced its partnership to use exaScreen® for novel drug discovery. The platform technology, developed by Pharmacelera, enables exponential screening capabilities of in-silico compound libraries, increasing the exploration of chemical space to identify hits. Traditional technologies can only mine about 10 million diverse molecules versus exaScreen®’s over 30 billion. The collaboration will focus on the application of exaScreen®’s state-of-the-art virtual screening technology for General Inception’s therapeutics companies looking for novel, diverse and synthesizable hits. exaScreen® uses accurate 3D Quantum-Mechanics (QM) molecular descriptors and Artificial Intelligence (AI) to mine efficiently a humongous chemical space and potentially boost the Intellectual Property (IP) of drug discovery projects.

“This is a step forward in our aim to retrieve and discover new chemical matter, which is a fundamental pillar in early drug discovery,” said Dr. Venkat Reddy, Chief Scientific Officer of General Inception. “Pharmacelera has already demonstrated that their proprietary technology is able to identify novel yet feasible scaffolds that are totally missed by traditional screening methodologies. We look forward to applying this technology and believe it will provide a meaning benefit to the drug discovery efforts of our companies.”

“We are very excited to establish this collaboration with General Inception, a fast-growing company Igniter in the United States and Europe,” said Dr. Enric Gibert, Pharmacelera’s CEO. “We appreciate getting in on the ground-floor of collaborations to best leverage the power of exaScreen®. GI’s business model enables us to reach a diverse group of innovative companies at the start of their journey and collaborate with pharma executives to resolve key challenges in drug discovery.”

ABOUT GENERAL INCEPTION

General Inception is pioneering company creation as an Igniter company. General Inception partners with extraordinary scientific founders at the inception of their journey to efficiently translate their groundbreaking innovations into transformational companies that address humanity’s grand challenges. As a business co-founder, GI brings together domain and functional expertise, executive talent, infrastructure and development resources, and capital to ignite, nurture and scale the company journey.

For more information, please visit www.generalinception.com

ABOUT PHARMACELERA

Pharmacelera develops advanced computational tools for the discovery of novel hits using accurate Quantum-Mechanics (QM), Artificial Intelligence (AI) and High-Performance Computing (HPC). The company’s products use 3D molecular descriptors to mine an unexplored chemical space and to identify hits uncovered by traditional algorithms. Pharmacelera is a private company founded in 2015 and based in Barcelona, Spain. The company works with several big pharma and biotech organizations across Europe and the United States.

For more information, please visit www.pharmacelera.com and www.pharmacelera.com/exascreen/

CONTACTS:

General Inception

Rebecca Galler – rebecca.galler@generalinception.com

Pharmacelera

Rémy Hoffmann – remy.hoffmann@pharmacelera.com

The post General Inception and Pharmacelera Partner to Advance Drug Discovery with Exponential Screening Capabilities appeared first on Pharmacelera | Pushing the limits of computational chemistry.

How Genetic Algorithms work?

Fernando Martín — Wed, 18 May 2022 07:29:21 +0000

Sometimes, the solution space in search and optimization problems cannot be expressed as a tree / graph of options (see previous AI post) or it cannot be easily traversed by heuristics or greedy algorithms. In those situations, we often use stochastic methods, those that heavily rely on a well described and rational random distribution. Genetic algorithms (or its broader concept of genetic programming) are one of such techniques.

Genetic algorithms are fascinating although they might look awkward at first. Do they really work? This is a question I have been asked many times. In fact, they do!

Genetic algorithms mimic the natural laws of evolution of living organisms that use genes as a way to code a solution to the problem of surviving in a specific environment. Such natural laws rely on natural selection and reproduction in a species to generate a population of best fit individuals.

For the sake of simplicity and clarity, we will use an example we described in our previous posts. Let’s imagine we want to design a molecule that best resembles a molecular reference by joining together 4 molecular building blocks from a pool of 65,536 building blocks. We will assume that any combination of building blocks is possible, that building blocks can be repeated, that the order of building blocks matters, and we will not take into consideration the chemical reactions required to link them.

Although the field has been widely studied and there are many aspects of genetic algorithms that can be solved differently (how to initialize the first generation or population, how to select individuals to cross over, how to mutate the individuals, how to measure the quality of the individuals, when to finish the search, …), we will use a few simple methods to understand the concept. In addition, most of these different methods depend on the particularity of the problem and they often need to be defined by human expertise.

In a genetic algorithm, an individual (or a solution to a problem) is digitally coded as a set of chromosomes, each chromosome representing a specific characteristic of the individual. In our particular case, each of the 4 building blocks of a potential molecule is a chromosome. Hence, a valid molecule (a valid solution to the problem of finding a molecule similar to a reference molecule) is coded as an array of 4 16-bit identifiers (we need 16 bits to code one of the 65,536 possible building blocks). Find below 3 examples of possible individuals (solutions to the problem).

We also need to define a fitness function, which describes how well the individual (the solution) satisfies the problem. In our case, we can use a Tanimoto-like metric to compare each molecule (each solution) to the reference molecule. We want to maximize this value.

Initially, the algorithm randomly generates a population of individuals (solutions). Let’s say a population of 10,000 randomly generated molecules (the identifiers of the building blocks are set to valid but arbitrary values). This population will become the first generation of the genetic algorithm and it will probably contain very poor individuals since their Tanimoto values will be very low (expected from a random population).

Next, the algorithm iterates over several generations in order to improve the individuals and to enhance the solution to the problem. At each iteration or generation, the algorithm selects individuals of the current population and it performs two different actions on them (cross over and mutation) to generate a similar size population (next generation) that will replace the previous one. The amount of iterations is very dependent on the problem. We can fix a number, we can define a convergence metric of the population or a combination of both (just in case the problem does not converge or it takes too long to do so). Let’s say we will iterate for 5,000 generations.

As commented, at each iteration, the algorithm selects pairs of two individuals, it crosses them over (reproduce) to generate a new individual, it might mutate it and it adds the new individual to the new generation (population). The iteration will be finished once the number of individuals in the next generation is equal to the number of individuals of the current generation.

In this select+cross+mutate cycle, the selection function prioritizes individuals from the current population that have a better fitness score. In this sense, it mimics natural selection by enabling good individuals (best fit) to have higher chances to generate off-spring (new solutions). The selection function also chooses poor individuals, although with less probability, to have a more diverse search space.

Crossing over (reproducing) two individuals means combining their chromosomes and generating a new individual (off-spring). Although several options exist, in our example, we will pick a random number of chromosomes from the first individual of the pair and the rest from the second. In the following figure, we show several potential offspring of two individuals.

Mutating an individual means randomly changing one or more of its chromosomes. Mutation is an action that needs to occur rarely but it will allow the algorithm to jump to a new part of the search space. In the following figure, we show several potential mutations of an individual.

At the end of all iterations, the best fit or the set of best fit individuals are the solutions that the algorithm has found.

The following pseudo-code depicts the behavior of the described algorithm. As a stochastic method, genetic algorithms are not deterministic in nature and will generate different solutions each time they are executed (there are mechanisms to make them “randomly deterministic” for debugging or reproducibility purposes).

Initialize random population P of X individuals
For a number of iterations do
  Create an empty new population P’
  Repeat until new population P’ has reached X individuals
    Select 2 individuals from current population P
    Cross over these 2 individuals to create a new individual
    Potentially mutate the chromosomes of the new individual with a low probability
    Add the new individual to the new population P’
  End repeat
  Make the new population P’ be the current population P
End for
Select the best individual/s as the pseudo-optimal solution/s

Graphically, the evolution of the population and the search for a solution would look like the following figures.

Initially, a randomly selected population will have individuals scattered around the search space as shown in the figure below, where the x-axis represents the solution space (each point being a different molecule or solution to the problem), the y-axis their Tanimoto value with respect to the reference (the value to maximize) and the circles the individuals of the population (for the sake of clarity we have just plotted 20 individuals, not 10,000).

Over time, since better individuals have more chances to be selected and crossed over, the population tends to the local / global maxima points, as shown in the figure below.

When the algorithm chooses to mutate an individual, it jumps to a new search space. If that new space is better than the current ones, this individual will have more probability to be selected and crossed over and the population will tend to move there. If the new space is worse, it does not have high chances to be selected and crossed over (low chances to survive). This is shown in the following picture for a single individual.

At the end, once the algorithm has iterated several times, the best individual or the best diverse individuals of the last population are the solutions found by the algorithm, as shown below. As commented before, the individuals of the last generation tend to be located near the local / global maxima points.

I hope this post has raised your interest in genetic algorithms and genetic programming and you will be eager to further read on this topic. Stay tuned for future Artificial Intelligence and Artificial Intelligence for Drug Discovery posts.

Want to learn more about how Pharmacelera can help you? Contact us and a member of our team will contact you to evaluate your project and how we can assist you with our technologies and services.

The post How Genetic Algorithms work? appeared first on Pharmacelera | Pushing the limits of computational chemistry.

Artificial Intelligence is NOT Machine Learning

Fernando Martín — Thu, 24 Feb 2022 05:51:38 +0000

By Enric Gibert – Feb. 24, 2022

Although the concepts of Artificial Intelligence (AI) and Machine Learning (ML) are used interchangeably, they are not the same.

Artificial Intelligence (AI) refers to the techniques and methods to demonstrate intelligence by machines, as opposed to the natural intelligence demonstrated by humans (supposedly ;-)). The field of AI was theorized in the 40s and 50s of last century. At a very high level, a machine would be considered intelligent if one interacted with it and could not differentiate whether machine replies were provided by another human sitting at the other side of the machine or by the machine itself. In other words, answers by a human and by a machine could not be distinguished.

AI is a very wide field that includes many areas such as expert systems, search algorithms, and Machine Learning (ML), among others.

Large knowledge-based databases such as expert systems consisting of a set of rules coded by humans is an example of AI. A toy expert system on vegetables can contain rules like “if round, green and size between 0.5 cm and 1.5 cm, then it is a pea”. However, one can imagine larger expert systems containing thousands of complex rules. Note that in this case, machines do not learn, as all knowledge is introduced (coded) by humans based on its expertise. Such systems seem intelligent, as they provide sound answers to specific questions. These expert systems can be equipped with inference algorithms to derive complex rules that are beyond the obvious human-coded rules. In this sense, inferencing is a first step towards machine learning.

Another AI area is search algorithms such as genetic algorithms, backtracking algorithms (in its depth-first, breadth-first or best-first versions such as A* search algorithms), or hill climbing. These algorithms try to find optimal solutions by exploring a wide range of potential candidate solutions. A chess game machine needs to decide what the best move is based on the current chessboard distribution and potential future moves by itself and by the opponent. The search space is humongous as the algorithm needs to evaluate and score many potential moves (“what if I move e4 – pawn to E4x -, the opponent moves e5 next – pawn E5 -, then I move NC3 – knight to C3 -, …” or “what if I move e4, the opponent moves e5 next, then I move BC4, …”) and chose the best one. The size of the search space depends on how deep (how many future movements) are considered. In this case, machines do not learn either, as they “simply” search for potential solutions that are built on-the-fly, but no knowledge or learning is retained or stored for future searches.

Machine Learning (ML) is the AI area in which machines automatically extract knowledge from data and experience that can be reused later to take decisions or propose predictions for new unseen situations. ML is being widely used in image recognition, speech recognition, web search algorithms, email filtering techniques, mating algorithms, … Regression analysis, artificial neural networks (and its more recent deep learning variants), support-vector machines and Bayesian networks, among others, are different ML techniques. In medical image recognition, for example, a deep neural network can be trained to early diagnose a particular neuro-degenerative disease using thousands of brain images. Once trained, the model can be used to predict whether new patients are developing that disease.

Evidently, sophisticated AI systems are hybrid systems that contain techniques from different areas. An algorithm for playing chess can contain human-typed rules (the most common chess openings, for instance), advanced backtracking search algorithms to decide the next move and a ML model to adapt the scoring function of the search algorithms based on previous actions by the opponent.

Since ML is the area of AI that is receiving more attention lately, these terms are often confused. But being ML a subset of AI, we can say that Artificial Intelligence is not Machine Learning, but Machine Learning is Artificial Intelligence.

Stay tuned for future short posts on Artificial Intelligence and Machine Learning.

The post Artificial Intelligence is NOT Machine Learning appeared first on Pharmacelera | Pushing the limits of computational chemistry.