Research

Presenting a new paradigm in AI-based drug discovery.

Early Discovery

Understanding disease mechanisms and unlocking new therapeutic possibilities

Our Early Discovery platform integrates multi-omics and large-scale clinical data analysis to quantitatively identify disease regulatory mechanisms and promising target candidates, conducting Translational Research that considers patient applicability. This AI-based target discovery infrastructure supports data-driven decision-making from target validation to lead discovery and preclinical strategy. Through an AI framework that integrates disease understanding, target discovery, and clinical connectivity into a single workflow, we enhance development direction and success probability from the early stages of drug development.

Core Capabilities

AI-driven Drug Target Discovery

Utilizing multi-omics data and AI-based prediction models to discover new therapeutic targets, prioritizing drug candidates considering clinical applicability by linking patient and clinical data.

System-level Understanding of Disease Mechanism

By integrating and analyzing large-scale multi-omics data, we quantitatively characterize molecular mechanisms of disease and system-level changes, and provide scientific evidence for establishing therapeutic strategies by linking these to clinical indicators and patient subtypes.

Neoantigen Discovery for Cancer Vaccine

Based on patient-derived omics data, we discover cancer cell-specific neoantigens and extend them to cancer vaccine development through immune response prediction and clinical linkage.

New Therapeutic Modalities

Understanding disease mechanisms and unlocking new therapeutic possibilities

This AI platform integrates chemical, sequence, and biological data at the Lead Screening and Optimization stages to propose novel candidates for various drug modalities and precisely improve their performance. By combining generative models with quantitative evaluation algorithms, it efficiently expands into design spaces difficult to explore with traditional experimental approaches, rapidly deriving promising Lead compound structures that comprehensively consider key development indicators such as efficacy, safety, and synthesizability. This reduces trial and error and uncertainty in the candidate discovery process, increasing clinical translation potential.

Core Capabilities

mRNA Sequence Optimization

In the vast sequence combination space corresponding to the 5'UTR–CDS–3'UTR regions of mRNA, we automatically explore and design mRNA sequences optimized for translation efficiency, stability, and immune response characteristics. Beyond initial models reflecting thermodynamic theory, search algorithms, and dynamic programming, we are currently advancing to next-generation mRNA sequence optimization models utilizing public data and deep learning architectures. Based on these technologies, we are building a generative platform capable of integrated design of all mRNA components including CDS, UTR regions, CAP, and polyA tail, aiming to significantly improve the speed and success probability of mRNA vaccine and therapeutic development.

mRNA–LNP Co-Design AI

We precisely optimize the composition ratio, physical properties, and delivery efficiency of LNP, which determines the effectiveness of mRNA therapeutics, through predictive and generative AI models. Through a co-design architecture that simultaneously considers the interaction between mRNA sequences and LNP composition, we propose novel LNP combinations with high tissue-specific delivery efficiency and excellent stability. This approach provides next-generation AI-based delivery vehicle design that optimizes the entire mRNA–LNP package as a single design space while reducing dependence on iterative experiments.

Antibody Engineering AI

Based on antibody sequence and structure information, we provide antibody generation and optimization models that comprehensively evaluate key developability indicators such as binding affinity, specificity, stability, and expression. By combining deep learning-based CDR design, structural stability improvement, and immunogenicity minimization algorithms, we derive high-quality antibody candidates with significantly improved efficiency compared to conventional sampling methods. This platform significantly improves the speed and accuracy of next-generation antibody design by precisely modeling antigen-antibody interactions.

Small Molecule & New Modality Generation

Using diffusion models, reinforcement learning (RL), and LLM-based generative models, we de novo generate and optimize various structures from small molecules to new mechanism modalities such as PROTACs and RNA binders. This platform comprehensively considers activity, toxicity, physical properties, and synthesizability to expand design spaces difficult to explore with existing methods and automatically proposes innovative structures. This provides generative AI that can rapidly build pipelines with new mechanisms beyond the limitations of traditional chemical design.

Translational Research

Understanding disease mechanisms and unlocking new therapeutic possibilities

We are building a preclinical and Translational Research platform that integrates Reliable ADMET, Optimized Design, and Biomarker Discovery into a single workflow, closely linking experimental data with AI analysis. This platform precisely predicts the pharmacokinetic properties and safety of candidate molecules, and through biomarker discovery, identifies patient responses in advance to increase clinical success probability.

Core Capabilities

Reliable ADMET

By applying Conformal Prediction techniques to a predictive architecture integrating LLM (Large Language Model) and graph-based models, we provide comprehensive ADMET profiles covering Absorption, Distribution, Metabolism, Excretion, and Toxicity of compounds along with confidence intervals. This enables quantitative management of prediction uncertainty at the candidate evaluation stage, supporting more accurate and reliable decision-making.

Molecule Design and Optimization

Our molecular design platform uses LLM and Diffusion model-based generative AI to learn patterns and rules of chemical structures, and designs and optimizes novel molecular structures expected to have therapeutic efficacy. By applying reinforcement learning to precisely improve molecules in the direction of satisfying target efficacy, pharmacokinetics, and safety characteristics, we more efficiently accelerate the derivation of innovative drug candidates.

Biomarker Discovery

By integrating and analyzing multi-omics data, clinical information, and real-world patient data (RWD), we systematically discover precision biomarkers that can be used for disease diagnosis, prognosis prediction, and drug response evaluation. Through AI-based analysis, we stratify patient populations and predict treatment responses, providing a precision medicine foundation that supports patient-tailored treatment strategy development and Companion Diagnostics development.

Reverse Translational Research

Understanding disease mechanisms and unlocking new therapeutic possibilities

Reverse Translational Research is a research approach that takes the rich patient data collected in clinical settings as a starting point, reinterpreting it at the basic research level to trace back the causes and mechanisms of disease. Through this approach, MOGAM Institute precisely analyzes patient electronic medical records (EMR) and various clinical data, discovering new research hypotheses based on symptoms and responses observed in actual patients. The mechanistic insights gained in this way are applied back to clinical decision-making, contributing to creating a virtuous cycle between research and clinical care.

Core Capabilities

EMR & Multi‑Omics Integration for Mechanism Exploration

To reveal the essential mechanisms of disease, it is necessary to integrate and analyze not only EMR but also various omics data including genomics, transcriptomics, proteomics, and metabolomics. This multi-omics integration approach is promising for elucidating pathological mechanisms, discovering biomarkers, and accelerating treatment development. Combining different omics layers enables more precise biological interpretation, patient stratification, and biomarker discovery, efficiently identifying disease-related genetic variations and metabolic signatures. Based on this, our institute finds new mechanistic clues in rare diseases and other complex conditions, establishing the scientific foundation for early diagnosis, prognosis prediction, and treatment strategy development.

Multi‑Agent-based Reasoning for Diagnosis, Treatment, and Patient Stratification Support

Recently, in the field of medical AI, multi-agent systems are gaining attention for extracting information from vast EMR and omics data and interpreting it effectively. Such systems have a structure where multiple specialized agents collaborate, each responsible for different roles including data collection and organization, diagnostic support, risk prediction, and treatment recommendation. MOGAM Institute is developing a multi-agent AI framework that supports EMR data standardization, multi-omics data analysis, new mechanism clue exploration, diagnosis, prognosis prediction, and treatment strategy derivation. Through this, we systematically manage clinical data and use it to validate research hypotheses and clinical judgments. Using this advanced technology, we perform EMR and multi-omics-based diagnosis and treatment development more rapidly and precisely, contributing to establishing customized strategies based on Patient Stratification when needed.