DNA-Encoded Libraries with ML-Supervised Data Processing
Explore the potential of DNA-encoded libraries, a method that gives access to screening up to 3B compounds in a single tube! Our services leverage the full potential of this approach, enabling efficient screening of diverse compounds, and streamlining the path to novel therapeutic breakthroughs.
Our DEL service can be delivered as a complete end-to-end solution or as individual modules tailored to your
project needs. We begin with experiment design and can support protein production if required. Several DEL
selections under the discuss conditions are performed, which can address selectivity to off-target proteins.
All the data is carefully decoded and undergoes comprehensive data analysis.
We offer flexible options for hit follow-up, including traditional off-DNA synthesis or an advanced DEL-ML
workflow that leverages machine learning and large chemical spaces to prioritize compounds. Final compound
synthesis, hit validation, and hit expansion can also be managed entirely through Chemspace, ensuring a
seamless and efficient discovery process.
At every stage of the project, we provide detailed reports and data summaries to keep you fully informed. We
also offer regular progress meetings and consultations, where all members of your team can be included to
participate in the decision making process.
Library options
- HitGen OpenDEL 4.0 – 3B compounds
- HitGen Open DEL 3.0 – 2B compounds
- AlphaMa DEL kit – 1B compounds
- Covalent DEL kit – 9M compounds
- Custom DELs
Types of screening
- Affinity selection
- Photo-crosslinking selection
- Covalent library screening
- Target RNA
Classical Workflow
The selected molecules are resynthesized off- or on-DNA for hit confirmation.
Advantages:
- The ability to rapidly screen large numbers of compounds in a single experiment at reduced cost compared to other approaches.
- High-quality data analysis allows for efficient identification of potential hits.
- Utilization of diverse DELs improves the chances to find hits for different target classes.
Advantages:
- The ability to rapidly screen large numbers of compounds in a single experiment at reduced cost compared to other approaches.
- High-quality data analysis allows for efficient identification of potential hits.
- Utilization of diverse DELs improves the chances to find hits for different target classes.
DEL-ML-CS Workflow
The data after the DEL screen is carefully processed, and this data is used for machine learning. This model can be applied to search in combinatorial chemical spaces up to 300B compounds (including Enamine REAL 77B and Chemspace Freedom 142B), utilizing our in-house algorithms.
Advantages over traditional DEL screening:
- Reduce off-DNA synthesis costs while gaining more chemical insights. On-demand compounds are more affordable and faster to obtain, allowing testing of up to 200 molecules for the price of 20 custom syntheses.
- Test more hypotheses by validating a greater number of chemical series at lower cost.
- Simplify hit expansion by combining analog search with ML-based scoring for faster, cost-effective optimization.
Advantages over traditional DEL screening:
- Reduce off-DNA synthesis costs while gaining more chemical insights. On-demand compounds are more affordable and faster to obtain, allowing testing of up to 200 molecules for the price of 20 custom syntheses.
- Test more hypotheses by validating a greater number of chemical series at lower cost.
- Simplify hit expansion by combining analog search with ML-based scoring for faster, cost-effective optimization.
What is a DNA-encoded Library?
A DNA-encoded chemical library (DEL) is a technology for the synthesis and screening of a large number of compounds aimed at accelerating the early stage of the drug discovery process. The main principle is to tag every molecule of the library with a unique DNA sequence, acting as a barcode, to identify a promising ligand for a biological target. Such an approach enables synthesis and screening up to 3 billion compounds in a single tube, thereby saving time and other resources.Application in Drug Discovery
DELs technology is widely used in modern drug discovery workflow for hit identification. Having access to a vast combinatorial library, with millions or even billions of compounds in a single test tube, allows us to screen potential drug candidates against a target much faster and more efficiently. Furthermore, the application of DELs in drug discovery significantly increases the chance of finding novel hits and optimizing them into a potent drug candidate.DEL Synthesis
For efficient DEL screening against the target of interest, we at Chemspace provide access to the in-stock library options, including:- HitGen OpenDEL 4.0
- 3B compounds HitGen Open DEL 3.0
- 2B compounds AlphaMa DEL kit
- 1B compounds Covalent DEL kit
- 9M compounds Custom DELs
DEL Screening
We provide different types of screening based on your project requirements:- Affinity selection. Classical is the most popular method of DEL screening.
- Photo-crosslinking selection. DEL screening approach when DNA-tagged compounds contain a photoreactive group capable of reacting with the target under UV light.
- Covalent library screening. Screening of covalent libraries is an effective strategy for covalent lead discovery. Covalent DELs consist of compounds that have covalent warheads, which, in turn, contribute to the covalent bond formation during screening.
-
Target RNA. Modern application of DELs to discover hit molecules that bind to RNA rather than proteins.
In all of these workflows, bound compounds are identified via amplification of their DNA tags by PCR, followed by next-generation sequencing (NGS).
Data Analysis
Our data analysis service for DNA-Encoded Library (DEL) selections is designed to help you efficiently identify and prioritize potential hits from even the most complex screening experiments. We process raw sequencing data from multiple selection conditions - such as different targets, protein mutants, or competitive binding setups - and perform normalization, enrichment analysis, and statistical validation to uncover meaningful binders. The result is a refined list of candidate molecules supported by quantitative metrics and visual reports, which will be synthesized on- or off-DNA for hit confirmation.Machine Learning Integration in DEL
DEL-derived datasets are a great source of large, structured datasets that can be labeled using statistical methods. This data serves as an ideal basis for Machine Learning (ML), while the models can be applied to external chemical collections - both in-stock and on-demand libraries.The combination of ML with DEL can be efficiently utilized for the hit identification stage to find starting points for optimization. The models can be also applied at the hit expansion stage for analogs prioritization (DEL-ML-based hit expansion).
Frequently Asked Questions
What distinguishes your DEL services from others?
Our DEL services offer a complete pipeline – starting with the experiment design all the way to hit
confirmation and hit expansion – all in one place. You get access to chemistry, biology,
cheminformatics, ML specialists and project management that can take your project from the earliest
stages to well-validated hits.
What types of targets are suitable for DEL screening?
DEL screening is ideal for purified proteins such as enzymes, receptors, and protein–protein
interaction domains. It can also be applied to membrane proteins and cell surface targets if binding
conditions are optimized. We will be happy to review your specific case and provide our
recommendations.
What is the typical turnaround time for a DEL project?
As our services are modular, you can benefit from its parts as well as the full pipeline. The exact
timelines will vary depending on the goal of the project.
How do you ensure data confidentiality and security?
Before starting our work, we are signing all the necessary documents with our potential customers to
ensure full confidentiality. If you have any additional requirements about security and data
transfer, we will be happy to accommodate them.
References
- McCloskey, K.; Sigel, E. A.; Kearnes, S.; Xue, L.; Tian, X.; Moccia, D.; Gikunju, D.; Bazzaz, S.; Chan, B.; Clark, M. A.; Cuozzo, J. W.; Guié, M.-A.; Guilinger, J. P.; Huguet, C.; Hupp, C. D.; Keefe, A. D.; Mulhern, C. J.; Zhang, Y.; Riley, P. Machine Learning on DNA-Encoded Libraries: A New Paradigm for Hit Finding. J. Med. Chem. 2020, 63 (16), 8857–8866. https://doi.org/10.1021/acs.jmedchem.0c00452