The Target 2035 initiative is transforming early drug discovery through open, collaborative protein science. Co-led by the SGC, this project advances open science by integrating protein production, high-throughput screening, and public data sharing. Central to this is the Protein Donation Program, enabling global contributions for ligand discovery screening. A co-developed roadmap outlines how FAIR data practices improve scalability, reproducibility, and access, accelerating hit discovery and target validation through shared infrastructure.

Artificial nanovesicles and extracellular vesicles need surface proteins to target cells and deliver drugs. Existing engineering takes weeks, handles few proteins, and yields mixed products. Here, we showcase EV-PRIME (EV-Protein Rapid Insertion by cell-free Membrane Engraftment). The one-pot, machine-learning-guided system uses cell-free synthesis to express and embed proteins onto vesicles within hours. It represents the first high-throughput, ML-directed platform for engineering protein-enhanced vesicles.

Cell free expression is a powerful technique for rapidly prototyping protein candidates in a discovery program. Gene templates are directly added to cell lysate yielding assayable quantities of proteins in a few hours. This talk covers our recent efforts on designing functional assays (protein binding and activity) that can be conducted directly in cell lysate, removing the need to purify the protein, thereby increasing data throughput for predictive models.

We present a novel antibody discovery approach enabling precise epitope specification and rapid timelines achieving sequence identification in 24 hours and KD determination within two weeks. Using the generative AI model Chai-2, we achieved a 16% hit rate in de novo antibody design, a 100-fold improvement over prior methods. In a single round, Chai-2 produced binders for 50% of 52 diverse targets, highlighting AI's transformative potential in biologics discovery.

This presentation describes a modeling strategy for antibody purification process fit assessment. Principal Component Analysis is applied to extract a one-dimensional basis for comparison of molecular chromatographic binding behavior from high-throughput screens. Ridge Regression is used to predict the principal component for new molecular sequences. This workflow is demonstrated with 97 monoclonal antibodies for five chromatography resins. Model development benchmarks four descriptor sets from biophysical descriptors and protein-language models.

To unravel the complex genetic grammar of protein expression, we developed a systematic approach for the construction of a gigantic parallel assay of combinatorial libraries, characterizing expression across hundreds of millions of genetic contexts. Leveraging a premade library of all E. coli regulatory elements, we generate high-resolution, ultra-high-throughput datasets of sequence-to-expression relationships. These data are then used for machine learning to predict and further optimize protein production, offering a powerful framework for data-driven protein engineering.

Viscosity is a crucial parameter for biotherapeutic development, but traditional measurements consume high volumes of sample (milligrams) for a single measurement of viscosity. To position viscosity screening earlier in the drug discovery pipeline, AbbVie developed the iBEACON, which measures curves of viscosity vs. concentration up to 150 mg/mL with a single experiment that consumes 100 micrograms of protein. This novel instrument allows us to collect data on ~10-fold more molecules per program than traditional approaches. Now, we are using these large data sets as the basis for building the next generation of predictive models of viscosity.

AI-driven protein and antibody design is constrained by slow, low-throughput experimental validation, and SPOC® overcomes this bottleneck by enabling on-chip synthesis and full kinetic screening of 192–1152 candidates to generate AI-ready, low-picomolar–resolution datasets in just 3–4 weeks. Using this platform, we profiled ~700 anti-HER2 scFv variants across multiple pH conditions, producing rich sequence–function kinetics matrices that directly support model training, active learning, and the rapid optimization of next-generation biologics.

This presentation will detail the discovery and engineering of VHH antibodies targeting the complex membrane protein CCR8, a Class A GPCR selectively expressed on tumor-infiltrating regulatory T cells. We will walk through the end-to-end workflow—from antigen preparation and immunization to panning, humanization, affinity maturation, and developability optimization—leveraging miniaturized screening assays to guide candidate selection. The talk culminates in the structural elucidation of the antibody-antigen complex, demonstrating the power of predictive approaches in membrane protein targeting.

We will discuss cell-free methods using various forms of nanodisc such as apolipoprotein, telodendrimers, and SMALPs to support and refold challenging membrane proteins, including large mammalian proteins over 200 kDa. This includes proteins like MOMP, CAR-T receptors, voltage-gated ion channels, and SARS-CoV-2 RBD that were all expressed in E. coli lysates and solubilized in synthetic or natural lipids. These approaches significantly enhance protein stability, solubility, and biological functionality, outperforming traditional refolding methods. Our strategy enables efficient production of therapeutic membrane proteins and supports new solutions for producing complex, high-molecular-weight mammalian proteins, addressing key challenges in protein biochemistry and biotechnology.

Selection systems such glutamine synthetase (Gs) and dihydrofolate reductase (Dhfr) have been used for decades to generate highly productive CHO cell lines. Using high-throughput CRISPR screens we identified asparaginase as a novel selectable marker that can be used alongside Gs in glutamine dropout media to generate cell lines with higher specific productivity and titer. Separately, we will share preliminary data on using essential amino acid biosynthesis as a selection system.