Fragment Library

In the research of covalent inhibitors targeting serine and threonine, scientists have found that the nucleophilicity of these hydroxyl groups is significantly enhanced due to the influence of their surrounding environment. This results in higher activity during catalytic reactions. Aspirin, which targets the non-catalytic domain serine (Ser529 in human COX1) of cyclooxygenase, exerts its anti-inflammatory effect through covalent binding. β-lactam antibiotics, which targets the catalytic domain serine of penicillin-binding proteins, interferes with bacterial cell wall synthesis.

Through careful selection, we constructed a structural filter containing over 110 electrophilic groups. By analyzing the electrophilic fragments selected by the structural filter, we removed any molecules with trivial or undesirable structural features. Ultimately, we obtained 3,300 fragment molecules which can target serine and threonine residues and can be used for fragment-based covalent drug discovery.

19F-NMR has proved to be a detection mode in fragment-based drug discovery (FBDD) for studies of protein structure and interactions. 19F shows high sensitivity for NMR detection, and the exquisite sensitivity of 19F chemical shifts and linewidths to ligand binding all make it a valuable approach in FBDD.F (Fluorine) -Fragments can be used for 19F-NMR detection after binding to target proteins, and can be used as an effective 19F-NMR tool for FBDD.

MCE designs a unique collection of 5,077 F-fragments, all of which obey a heuristic rule called the “Rule of Three (RO3)”, in which molecular weight ≤300 Da, the number of hydrogen bond donors (H-donors) ≤3, the number of hydrogen bond acceptors (H-acceptors) is ≤3 and cLogP is ≤3. This F-fragments library is an important source of lead-like drugs.

A diverse compound library with favorable ADMET properties (Absorption, Distribution, Metabolism, Excretion, and Toxicity) is crucial in drug discovery. Early evaluation of ADMET properties allows for the exclusion of molecules with unfavorable profiles at the initial stages, thereby reducing the risk of late-stage development failures, lowering R&D costs, and accelerating optimization of lead compounds. Based on predictions from ADMET-related AI algorithms, the compounds in this library are predicted to exhibit favorable oral bioavailability (F > 30%), reasonable plasma protein binding (PPB < 98%), minimized CYP3A4 inhibition potential (inhibition probability < 50%, CYP3A4 is the most critical drug-metabolizing enzyme in the cytochrome P450 family) , low toxicity profiles, with 140 potentially toxic substructures pre-identified and excluded via substructure searching to eliminate compounds containing hazardous fragments. The diversity library enables broad applicability in high-throughput screening (HTS) and high-content screening (HCS).

CRBN, namely cereblon, is the substrate recognition subunit of the E3 ubiquitin ligase complex in the ubiquitin-proteasome system. A CRBN ligand library refers to a collection of numerous fragments that can specifically bind to the CRBN protein.

These ligands are mostly designed based on validated CRBN-binding warheads and modified through AI-driven molecular generation optimization systems. They not only include classic lenalidomide-derived structures but also cover novel non-lenalidomide scaffolds. After drug-likeness filtering, these ligands exhibit structural diversity and favorable druggable properties. They can be further optimized and modified to facilitate the development of novel molecular glue degraders, accelerate the discovery of molecular glues that induce interactions between CRBN and new substrate proteins, and enable the exploration of novel CRBN substrates for identifying previously unknown CRBN-binding proteins.

MCE compiles 122 fragments that can specifically bind to the CRBN protein, with molecular weights ranging from 200 to 500. Compounds developed based on the library ligands target multiple disease targets such as cancer and autoimmune diseases, further advancing the development of Molecular Glues and PROTACs therapeutic agents.

The discovery of hit molecule is a cornerstone of drug development. Among the diverse tools available, DNA-encoded libraries have emerged a revolutionary platform for high-throughput screening. Compared with traditional HTS, DEL features shorter screening processes, lower costs, simpler assays, and larger library capacities.

DEL Construction utilizes split-and-pool synthesis, a combinatorial chemistry approach that involves iterative splitting, reaction, and pooling. This strategy enables rapid, exponential assembly of fragments in minimal steps without the need for individual compound synthesis andassoicicated isolation or purification steps, thus greatly reducing overall costs. The technology enables simultaneous affinity screeningof massive compound collections to target proteins in a single step. By coupling chemical structures with unique DNA barcodes, each compound is tagged with a distinct DNA sequence for convenient tracking and decoding.DELs readily enable the construction and efficient screening of libraries containing millions to billions of compounds. As a result, DEL screening combines the dual advantages of high efficiency and low cost, making DEL a transformative technology in modern drug discovery.

The DEL kit consists of 50 independent libraries with a total scale of 100 billion compounds. It is constructed through stepwise combinatorial chemistry strategies involving 2-, 3-, and 4-round synthesis. By employing diverse scaffolds and flexible linking strategies, it encompasses various ring systems, linear frameworks, and heterocyclic structures. Screening can be achieved solely through affinity, independent of target-specific activity detection methods. This library is suitable for DEL screening against a wide range of targets.

Protein protein interactions (PPI) have pivotal roles in life processes. The studies showed that aberrant PPI are associated with various diseases. However, the design of modulators targeting PPI still faces tremendous challenges, such the difficult PPI interfaces for the drug design, lack of ligands reference, lack of guidance rules for the PPI modulators development and high-resolution PPI proteins structures.

The PPI Library comprises molecules of various sizes, frameworks, and shapes ranging from fragment-like entities to macrocyclic derivatives designed as secondary structure mimetics or as epitope mimetics. The designs cover β-turn / loop mimetics and α-helix mimetics. Since helices present at the interface in 62% of all protein-protein interactions. This library focused on designs including mimics with the substitution geometry of an a-helices, as well as designs that mimic the location of “hot-spot” side chains in helix-mediated PPIs.

Small molecule covalent inhibitors, or irreversible inhibitors, are a type of inhibitors that exert their biological functions by irreversibly binding to target through covalent bonds. Compared with non-covalent inhibitors, covalent inhibitors have obvious advantages in bioactivity, such that covalent warheads can target rare residues of a particular target protein, thus leading to the development of highly selective inhibitors and achieving a more complete and continued target occupancy in living systems. In recent years, the distinct strengths of covalent inhibitors in overcoming drug resistance had been recognized. However, toxicity can be a real challenge related to this class of therapeutics due to their potential for off-target reactivity and has led to these drugs being disfavored as a drug class. The drug design and optimization of covalent inhibitors has become a hot spot in drug discovery.

MCE covalent inhibitor library contains 6,121 small molecules including identified covalent inhibitors and other molecules having common covalent reactive groups as warheads, such as acrylamides, activated terminal acetylenes, sulfonyl fluorides/esters, cloracetamides, alkyl halides, epoxides, aziridines, disulfides, etc.

MCE Covalent inhibitor Library plus, with more powerful screening capability, further complement Covalent inhibitor Library (HY-L036) by adding some fragment compounds with covalent warheads.

Macrocyclic compounds (≥12-atom cyclic small molecules/peptides) have unique physicochemical properties. They form preorganized conformations with high binding affinity/selectivity, target traditional small-molecule-inaccessible proteins, and bridge small-molecule drugs and biological agents. As key protein phosphorylation enzymes, kinases are linked to tumors, COPD, etc., and are critical therapeutic targets. Traditional small-molecule kinase inhibitors lack selectivity, causing off-target toxicity, low bioavailability, and acquired resistance. Macrocycles’ semi-rigid structure restricts conformations, boosts binding selectivity, optimizes pharmacokinetics, and makes macrocyclization a core kinase inhibitor optimization strategy.

Thousands of bioactive macrocycles were curated from ChEMBL. Via Transformer, macrocyclization was converted into a chemical language translation task, enabling end-to-end macrocycle generation from linear precursors with simplified inputs. Macformer achieves efficient, automated linear molecule macrocyclization via deep learning; generated macrocycles have diversity, novelty, biocompatibility, and cover broader chemical space.

MCE collected thousands of marketed/clinical kinase inhibitors, using their fragments for macrocyclization to generate derivatives. After evaluating synthetic accessibility and physicochemical properties, a million-scale virtual macrocyclic library was built for kinase-related virtual and AI-driven screening.

Macrocyclic compounds (≥12-atom cyclic small molecules/peptides) have unique physicochemical properties. They form preorganized conformations with high binding affinity/selectivity, target traditional small-molecule-inaccessible proteins, and bridge small-molecule drugs and biological agents. As key protein phosphorylation enzymes, kinases are linked to tumors, COPD, etc., and are critical therapeutic targets. Traditional small-molecule kinase inhibitors lack selectivity, causing off-target toxicity, low bioavailability, and acquired resistance. Macrocycles’ semi-rigid structure restricts conformations, boosts binding selectivity, optimizes pharmacokinetics, and makes macrocyclization a core kinase inhibitor optimization strategy.

Thousands of bioactive macrocycles were curated from ChEMBL. Via Transformer, macrocyclization was converted into a chemical language translation task, enabling end-to-end macrocycle generation from linear precursors with simplified inputs. Macformer achieves efficient, automated linear molecule macrocyclization via deep learning; generated macrocycles have diversity, novelty, biocompatibility, and cover broader chemical space.

MCE collected thousands of marketed/clinical kinase inhibitors, using their fragments for macrocyclization to generate derivatives. After evaluating synthetic accessibility and physicochemical properties, a million-scale virtual macrocyclic library was built for kinase-related virtual and AI-driven screening.