How Can We Help You with Your Projects?

Contact us to discuss the specifics of your project. Our team of experienced scientists is ready to assist you with tailored solutions to meet your unique needs. Whether you need help with assay development, FDA IND, stability study, lot release approval, or other scientific endeavors, we are here to support you.Contact us to learn more about our cell based assay services

How Long Does It Take to Get Started?

We will initiate your project as soon as the proposal is approved and signed by you. Our streamlined process ensures a quick start, minimizing delays and maximizing efficiency. Fill out the form below to email us or click our live chat option to speak with a MarinBio Scientist and get your project underway.

Can You Help with Assay Development and Validation for FDA IND, Lot Release and Stability Study Approval?

Yes, we have over 30 years of expertise in developing and validating cell based assays for FDA IND, lot release or stability study approval. Our team can also liaise directly with the FDA on your behalf, ensuring a smooth and compliant process. Please contact us to learn more about how we can assist you with your assay development and FDA approval needs.

Do You Need a Custom or Tailored Assay?

MarinBio scientists are experienced and creative in designing, developing, validating assays. Some of the assays we perform are listed below, but there are many more that we created, developed and validated. Give us call so that we may help you.

AI Driving the Future of Drug Discovery

Download the full article (PDF)

Generative Intelligence in Therapeutics: A New Framework for Designing Small Molecules, Proteins, and Emerging Modalities

Artificial Intelligence has progressed from a supplementary analytics tool to a foundational infrastructure that now supports the full drug discovery and development spectrum. Recent, rapid advances in generative models, multimodal transformers, and molecular simulation engines, have dramatically redefined the tempo, accuracy, and feasibility of therapeutic design. Definitions

Generative models: AI algorithms that design new molecular structures by learning patterns from existing chemical and biological data.
Multimodal transformers: Neural network architectures that integrate multiple data types, such as protein sequences, molecular structures, and assay results, to predict properties or interactions.
Molecular simulation engines: Computational platforms that model the physical and chemical behavior of molecules to predict stability, binding, or dynamics in silico.

AI is no longer positioned as an experimental approach but is recognized as a core strategic differentiator across pharmaceutical and biotechnology enterprises. AI-native companies have shown that computational design can yield viable preclinical candidates in a fraction of traditional timelines, while established pharmaceutical organizations increasingly rely on AI-driven discovery strategies to expand their modality toolkits, especially into targets long regarded as undruggable.

AI-Enabled Drug Discovery Modalities

AI-enabled drug discovery spans multiple therapeutic classes, each influenced by unique datasets, structural constraints, and design objectives. At present four major modalities dominate both industry investment and scientific discourse. The two most advanced modalities are: AI-enabled small-molecule discovery and AI-designed protein therapeutics. New approaches are also being heavily impacted are AI-guided targeted protein degradation modalities, and AI-designed peptide, and peptidomimetics. These modalities differ in molecular size, physicochemical properties, designability, and regulatory precedents, yet they increasingly share a common computational foundation. Advances in sequence-to-structure modeling, molecular generative modeling, and large experimental datasets have enabled cross‑modality learning that accelerates discovery across each category. Table 1. Comparison of Leading AI-Enabled Drug Modalities

Modality	Molecular Type	Strengths	Challenges	Maturity Level
AI-enabled small molecules	3D molecular ligands	Strong regulatory precedent; fast optimization	ADMET complexity	Most mature
AI-designed biologics & antibodies	Proteins, antibodies, decoys, scaffolds	High specificity; programmable binding; undruggable target access	Manufacturability, immunogenicity, long timelines	High momentum
AI-guided degraders	PROTACs, molecular glues	Event-driven pharmacology; eliminates proteins	Size, permeability	Emerging
AI-designed peptides & peptidomimetics	Peptides, pseudo-peptides	Protein-protein targeting; high specificity	Stability, delivery	Early-stage

AI-Enabled Small Molecule Discovery

Small molecules remain the most regulatorily familiar and clinically validated therapeutic class. AI tools now enable rapid exploration of the chemical universe, with machine-learning–guided compound design tools producing active molecules. Design of small molecules are guided by 3D structural constraints using integration of physics-based simulation, ADMET prediction (see next paragraph), and synthetic accessibility scoring. Definitions

3D structural constraints: Ensure that AI-generated molecules adopt physically realistic geometries that match the steric, spatial, electrostatic, and conformational requirements of the target binding site, increasing the likelihood of true binding.
Integration of physics-based simulation: Combines AI-generated designs with molecular dynamics, free-energy methods, and quantum-chemical calculations to refine candidates and provide high-fidelity estimates of binding affinity, stability, and conformational behavior.
ADMET prediction: Predicts absorption, distribution, metabolism, excretion, and toxicity properties early in design to prioritize drug-like candidates and filter out liabilities such as poor solubility, metabolic instability, or toxicity.
Synthetic accessibility scoring: Evaluates how feasible a molecule is to synthesize, based on retrosynthetic complexity, functional-group compatibility, and known reaction rules, so that AI proposes compounds that are both potent and practically manufacturable.

Table 2. Examples of AI-designed small-molecule drugs currently in clinical trials

Drug (company)	How AI assisted in design	Target & mode of inhibition	Disease indication	Clinical phase
DSP-0038 (Exscientia)	AI-driven design/lead optimization to generate desired receptor profile and drug-like properties; accelerated timeline to first-in-human.	Serotonin receptor modulator, reported as 5-HT₁A agonist / 5-HT₂A antagonist	Alzheimer’s disease psychosis	Phase 1 (entered in/around 2021).
ISM001-055 / Rentosertib (Insilico Medicine)	Generative AI for de-novo target identification and to design novel small-molecule chemotypes; AI accelerated hit-to-lead and candidate selection leading to rapid progression into clinic.	TNIK (TRAF2- and NCK-interacting kinase) small-molecule kinase inhibitor	Idiopathic pulmonary fibrosis (IPF).	Phase 2a (positive Phase 2a reported)
BEN-8744 (BenevolentAI)	AI platform helped nominate a novel target/chemotype and supported design/prioritization of small molecules (multi-data integration for target to chemistry).	PDE10A developed as a peripherally-restricted PDE10 inhibitor (oral small molecule).	Ulcerative colitis / inflammatory bowel disease (IBD).	Phase 1 / Phase 1a completed

Four Reasons Why AI-Designed Small Molecules Are Under Intense Investigation:

GENERATIVE AI. In small-molecule drug discovery, Generative AI refers to a class of artificial intelligence models, usually deep learning architectures, that create new chemical structures rather than simply analyze existing ones. Unlike traditional computational approaches that screen or optimize predefined libraries, generative AI invents novel molecules that fit desired biological, physicochemical, and developability constraints. Generative AI is dramatically accelerating “design-make-test” cycles. New tools can propose novel chemical structures, optimize binding affinity and drug-like properties, and even suggest synthetically feasible molecules, often orders of magnitude faster than traditional medicinal chemistry workflows. CLINICAL TRIALS. The first AI-designed small molecule drugs are now in (or nearing) clinical trials. For instance, Rentosertib was developed via a generative-AI platform and targets TNIK for fibrosis. Its progress through early clinical stages marks a major milestone validating AI-driven small-molecule design. BROAD APPLICABILITY AND SCALABILITY. Small molecules remain the backbone of most current drugs. AI excels at scanning, or generating a wide breadth of chemical structures, and predicting ADMET (absorption, distribution, metabolism, excretion, toxicity) properties. Because of this, AI-enhanced small-molecule discovery remains highly attractive to both established pharma companies and startups. EXPANDING CAPABILITIES. Targeting “undruggable” proteins, proteolysis, and more. AI isn’t just producing standard inhibitory molecules. There is growing work designing small molecules that act via non-traditional modalities, e.g. small-molecule degraders, allosteric inhibitors, or ligands for previously intractable targets. Because of these developments, many in the field view AI-enabled small-molecule discovery as “low-hanging fruit”, high impact, relatively well-understood chemistry, and a high likelihood of success in near-term drug pipelines.

AI-Designed Protein Therapeutics (Antibodies & Biologics)

AI-designed protein therapeutics have emerged as one of the fastest-growing application areas in drug discovery, driven by major advances in protein structure prediction and generative sequence modeling. Unlike classical antibody discovery, which relies on natural immune repertoires, AI systems can now produce entirely synthetic antibody or protein sequences that fold into predetermined shapes engineered for target-specific interactions.For example, AI can generate proteins with high stability, and optimized paratope-epitope complementarity. Paratope-epitope complementarity refers to the precise, three-dimensional and chemical fit between the paratope (the antigen-binding surface on the antibody) and the epitope (the specific molecular feature on the target antigen that the antibody recognizes and binds). Paratope-epitope complementarity describes how well the antibody’s binding site matches the shape, charge distribution, hydrophobic/ hydrophilic patterning, and conformational features of the antigen’s epitope. High complementarity enables strong, specific, and stable binding, much like a finely matched lock and key, but with more flexibility. This capability allows researchers to target difficult classes such as GPCRs, ion channels, and multi-domain receptors. Additionally, AI enables rational multispecificity design, permitting simultaneous engagement of two or more pathogenic targets or epitopes. Biologics / Protein-based Therapeutics: the fast-rising second wave While small molecules get a lot of attention today, biologics and protein therapeutics designed or optimized with AI are rapidly gaining traction: AI-driven platforms are now being used to engineer antibodies, protein scaffolds, and other biologics with improved binding, stability, and manufacturability. Several companies are explicitly partnering for “protein-based therapeutics from scratch,” using AI to design proteins de novo in weeks rather than years. For complex diseases (e.g. oncology, immunology) where traditional small molecules fail, AI-designed biologics may offer breakthroughs, especially for targets considered “undruggable” by small molecules. Because biologics often involve more complex chemistry/biology and manufacturing challenges, the maturation and regulatory path tend to be longer. But many experts believe biologics, especially AI-optimized ones represent the “next frontier.”

Cell-Based Assays are a Bottleneck for Testing AI-Enabled Drug Modalities

Cell-based assays are currently one of the major bottlenecks in validating AI-driven drug designs. The primary challenge arises from the throughput gap between computational prediction and experimental validation. Cellular assays typically operate at low-to-medium throughput due to limitations in transfection efficiency, expression, and automation capacity. Consequently, only a small fraction of predicted proteins can be experimentally evaluated. Biological complexity further compounds the issue. Cellular assays introduce context-dependent effects, differences in folding, post-translational modification, or trafficking that are not captured by computational models. Even identical proteins may behave differently across cell lines, complicating reproducibility and interpretation. Scaling also presents challenges: variability in culture conditions, passage number, and assay sensitivity can generate inconsistent results, particularly in partially automated settings. Cost and infrastructure remain significant barriers. Large-scale cell-based assays require robotics, automated microscopy, and sophisticated data management systems, whereas AI-based prediction is computationally inexpensive once established. Moreover, AI models require extensive, labeled experimental datasets for retraining and refinement, yet current cell-based validation throughput limits data generation and slows feedback loops. A systematic approach, combining mechanistic insight, validated resources, physiological relevance and cell-based assay selection becomes a critical bridge linking computational design to biological activity.

Conclusion

The integration of AI into drug discovery has transitioned from a promising experimental concept to a transformative strategic capability that reshapes the pharmaceutical landscape. By enabling rapid generation of viable preclinical candidates and facilitating exploration of previously intractable targets, AI is accelerating the pace of innovation across multiple therapeutic modalities. The convergence of computational techniques, ranging from sequence-to-structure modeling to molecular generative algorithms, provides a unifying framework that transcends differences in molecular class, size, and complexity. This shared computational foundation not only enhances the efficiency and precision of small-molecule and protein therapeutic design but also catalyzes innovation in emerging modalities such as targeted protein degradation and de novo peptide therapeutics. As AI-driven platforms continue to evolve, they are poised to redefine the traditional boundaries of drug discovery, reduce attrition in preclinical development, and ultimately expand the universe of treatable diseases. The ongoing cross‑modality learning and data integration underscore that AI is not merely a tool but a critical enabler of a new era in therapeutic development, where computational insight and experimental validation operate in seamless synergy. Disclaimer: This blog post is intended solely for educational and scientific informational purposes. Any mention of therapeutic drug names, including FDA-approved medications, is for the purpose of accurate reporting and discussion of biomedical research and does not constitute medical advice, endorsement, or promotion. Readers should not interpret the content as a recommendation for any specific treatment. Always consult a qualified healthcare professional for medical advice or treatment decisions.

About Marin Biologic Laboratories

Our Recent Publication/Meeting Presentation on Gene Therapy

1. Development of VNX-101, an Adeno-Associated Virus with Less Immunogenicity and Efficient Long-Term Expression of a CD19 T-Cell Engager. Molecular Therapy Methods & Clinical Development, published online July 24, 2025.

3. Cell-Based Potency Assay for Anti-CD3-Anti-CD19 Diabody. Journal of Immunological Methods. 2025. 545-114004.

3. Development of a Pharmacokinetic (PK) Mouse Serum GLP ELISA for an Anti–CD19–AntiCD3 Diabody bioRxiv 2025.03.19.644217; doi: https://doi.org/10.1101/2025.03.19.644217.

4. American Society of Hematology (ASH) Annual Meeting 2024. Abstract link: Using Gene Therapy to Solve Challenges with CAR-T Cell Immunotherapy: Lead Selection and Preclinical Development of an Adeno-Associated Virus with Reduced Immunogenicity Exhibiting Efficient and Long-Term Expression of an Anti-CD19 T-Cell Engager.

Comprehensive Assay Solutions for In Vitro and Cell Based Potency Assays and Pharmacokinetics Studies- Our Expertise

With 30 years of expertise in cell culture, cell-based assays, and preclinical/clinical PK/PD analysis, we specialize in offering assay services essential for a wide variety of therapeutic drug development programs, preclinical studies, IND/BLA applications, and commercialization. Our comprehensive services include both preclinical non-GLP and GLP assays, as well as non-GMP and GMP assays, providing critical support throughout the entire development pipeline. Watch the following video and explore our latest presentation on the development and validation of potency and pharmacokinetic (PK) assays for AAV vectors, highlighting innovative methodologies and industry-leading expertise.