Company Platform Pipeline Contact us

Sapian is Kantify's AI-based drug discovery platform

Sapian is Kantify's Artificial Intelligence platform for drug discovery. Sapian is a proprietary platform, built in-house, and is made up of multiple applications that each solve one or more fundamental challenges involved in the drug development process.

Discover some of the capabilities of Sapian, and example use cases of the technology below.

Sapian.Target finds novel drug targets

The Problem
Most drug targets are still unknown
Discovering which drug targets are involved in diseases is a fundamental unsolved challenge in drug discovery. Studies show that fewer than 700 human proteins are the putative target of FDA approved drugs - a tiny fraction of the ~100,000 protein variants thought to exist, not to mention other druggable biomolecules, such as RNA. To add insult to injury, some drug targets only seem to have a biological effect when modulated in combination with other targets - causing an exponential explosion in possible choices of targets to validate. With the current techniques available to identify drug targets, which are slow, expensive, and error-prone, we won't make the progress needed to significantly accelerate the discovery of effective drugs without significant breakthroughs.
Our approach
We are creating a (ranked) atlas of all druggable targets
Sapian.Target discovers druggable protein targets involved in a biological process or disease. Our approach is literature-agnostic, and machine-learning based. Where relevant, we complement these predictions with "Omics" data and a variety of literature sources to narrow down on targets of interest. The end result is a comprehensive atlas of drug targets and their involvement in diseases. We've validated Sapian.Target extensively, and have already used this algorithm to discover a variety of novel drug targets in multiple diseases.
Patients affected by different forms of Limb-Girdle Muscular Dystrophies. From left to right: R1, R2, R4, R7 and R9

Target Discovery in Limb-Girdle Muscular Dystrophies

A use case of Sapian.Target
Therapeutic Area Neuromuscular disorders
Partner I-Stem
Discovery 2024

Limb-girdle muscular dystrophies (LGMD) are a group of rare genetic muscle disorders. LGMD R2, a subtype of LGMDs, is caused by a loss of the function of dysferlin, a protein that plays a key role in repair mechanisms in skeletal muscles. Patients affected by this disease will generally progressively lose mobility throughout their lives. Like most neuromuscular disorders, no approved treatment exists today.

In an effort to advance the understanding of the disease, we used Sapian.Target to identify multiple novel, druggable targets for LGMD R2. We validated the potential of these targets in multiple ways: firstly, by silencing the targets (using siRNAs) in a cellular model of the disease (based on induced pluripotent stem cells - iPSCs), we showed that LGMD R2 cells were particularly sensitive to the presence of these targets to maintain their structural integrity. Secondly, we identified that multiple repurposable drug candidates for LGMD R2 are dependent on the presence of these targets to induce their positive effect - further validating the targets' importance in treating LGMD R2, and also demonstrating novel Mechanisms of Action (MoA) through which repurposable compounds act to induce their effect.

This work has opened up the road to develop targeted therapies for LGMD patients. Furthermore, we can now start the work to identify other disorders that could be treated using this MoA - resulting in a potential far beyond individual rare diseases.

Sapian.Hit finds promising compounds

The Problem
There are too many chemical compounds to test experimentally
The experimental evaluation of a chemical compound is a slow and costly affair. Beyond the synthesis of the compound, one needs to design a proper assay, set up controls, and often run multiple replicates of an experiment to get stable results. Researchers have attempted to tame this complexity through high-throughput screenings, which involve using specialized machines to evaluate large, existing compound libraries on a well-standardized assays. Once set up, the largest of these screens can screen hundreds of thousands of compounds in just a few weeks. While these techniques have been pivotal in advancing drug discovery, they only allow us to uncover a minuscule fraction of the whole biochemical space in which life-saving molecules are hiding.
Our approach
We are building an algorithm that understands all compound/protein interactions
Sapian.Hit discovers compounds that interact with our targets of interest. We've taken an orthogonal approach to existing computational hit discovery algorithms, and we now have evidence that Sapian.Hit generalizes over the whole eukaryotic proteome and an extremely large chemical space - meaning it is capable of finding drugs for most diseases humanity faces. We're using Sapian.Hit on a daily basis for all of our projects, and have already found multiple first-in-class drugs with it which we're further developing.
A CT and PET scan of a patient with Metastatic Castration-Resistant Prostate Cancer. Red arrows highlight metastases.

New drugs for Metastatic Castration-Resistant Prostate Cancer

A use case of Sapian.Hit
Therapeutic Area Oncology
Partner Johns Hopkins University
Discovery 2023

Metastatic Castration-Resistant Prostate Cancer (MCRPC) are aggressive cancers that arise in response to first-line hormone therapy in Prostate Cancer. Unlike Prostate Cancer, which generally has a relatively good prognosis, MCRPC are extremely aggressive cancers, which very high recurrence rates and high tendency to further metastize. MCRPC does not have adequate treatment, meaning most patients diagnosed with this type of cancer have very poor outcomes.

In an effort to advance treatment of MCRPC, we used Sapian.Hit to identify New Chemical Entities that acted on novel targets we had previously identified. We virtually screened close to 10 million compounds in order to shortlist 3 candidates for wet-lab testing. Two compounds showed significant target engagement and significantly reduced proliferation in cellular assays. Our best compound was tested in vivo, where it induced full growth arrest on tumors, without exhibiting any notable toxicity.

This drug is currently undergoing further development in order to advance towards human trials.

Sapian.Ward makes our drugs safe and effective

The Problem
Not all compounds make good drugs
Drugs interact with our bodies in many surprising ways. Whether administered orally, topically, through an injection, or any other method, most drugs will interact with multiple organs and biological systems. These can block, transform, or help to clear drugs. Furthermore, inside our bodies, drugs can interact with other drugs or have unintended effects. Figuring out which drugs are going to be safe and effective is a tremendously challenging problem. Because of high failure rates, the very first step of clinical trials of novel drugs involves evaluating the safety of drugs when administered in humans. We believe we should aim for a near 100% success rate here, and AI is one tool to achieve this.
Our approach
We are predicting how compounds act in our bodies
Sapian.Ward predicts how a drug will behave in a broader biological context. We predict a wide variety of properties related to Absorption, Distribution, Metabolism, Excretion and Toxicology (ADMETox) of compounds, in order to shortlist compounds that are the most likely to be both safe and effective in humans. As a consequence, we filter out compounds that are most likely to fail, either in animal tests or in human tests, which further reduces the number of compounds we test in vivo and speeds up our discoveries.
Heart cells of a healthy rat (left) versus those having undergone Doxorubicin-induced cardiotoxicity (left).

Preventing Cardiotoxicity

A use case of Sapian.Ward
Therapeutic Area Ion Channels
Partner Universite de Nantes
Discovery 2023

Cardiotoxicity is a common cause for failure of drugs, with estimates ranging from 5%-15% of drug pipeline failing because of unacceptable heart toxicity. Often, this toxicity is caused by a drug inhibiting the function of a particular protein called Kv11.1, encoded by the hERG gene. Drugs that interact with this protein can cause various arrythmias or cardiac arrest, and sadly, it requires specialized assays or advanced animal tests to uncover this toxicity, often abbreviated in hERG toxicity.

We used Sapian.ward to detect whether we could differentiate between small molecules that selectively interacted with multiple promising targets that are closely related to hERG, and those that inhibited both hERG and our target of interest. We tested over 200 compounds from different chemical scaffolds, and purposefully included an equal amount of likely hERG-toxic compounds to be able to validate that experimental results matched with our predictions.

We showed that we achieve a near 95% (balanced) accuracy in predicting hERG toxicity on unseen compounds with very high chemical diversity, further validating that we can indeed identify, and filter out, compounds that cause this major issue in drug development.