Development of
a hit-calling algorithm on DEL selection data

Overview

DNA-encoded libraries (DEL) have become a powerful approach for identifying potential drug candidates in early-stage drug discovery. However, analyzing DEL sequencing results requires sophisticated computational workflows capable of processing large volumes of next-generation sequencing (NGS) data while accurately identifying meaningful compound hits.

Excelra collaborated with a European pharmaceutical company to develop a hit-calling algorithm on DEL selection data that could reliably identify candidate molecules from large sequencing datasets. By combining advanced Bioinformatics Solutions, robust Scientific Data Management, and scalable computational pipeline platforms, Excelra created an automated workflow capable of processing high-throughput DEL datasets and improving hit identification for drug discovery.

Our client

The client is a European pharmaceutical company dedicated to discovering and developing small-molecule therapeutics with novel mechanisms of action. Their research team required advanced computational tools to analyze large DNA-encoded library datasets and identify promising compounds for further testing.

Client’s challenge

DNA-encoded library experiments generate massive volumes of sequencing data that must be carefully analyzed to identify meaningful compound hits.

The client faced several challenges:

• Processing large volumes of next-generation sequencing (NGS) data
• Limited compound coverage within DEL datasets
• High noise levels in sequencing data affecting hit detection
• Difficulty distinguishing true hits from false positives
• Need for a scalable algorithm integrated with high-throughput drug discovery workflows

To address these challenges, the client required a statistical approach capable of improving hit identification accuracy while handling noisy datasets.

Client’s goals

The project aimed to:

• Develop a hit-calling algorithm for DEL selection data
• Improve detection of true compound hits
• Reduce false positives in sequencing analysis
• Integrate DEL analysis into automated computational pipelines
• Enable scalable NGS data processing for drug discovery workflows

Our approach

DEL dataset analysis

Excelra began the project with a comprehensive bioinformatics analysis of the DEL dataset. The dataset consisted of 72 samples containing raw count data for approximately 5 million compounds in the client’s library.

Because low raw count values increased the risk of false positives or missed hits, data normalization was critical for accurate hit identification.

Data standardization and normalization

Excelra implemented data standardization techniques and statistical normalization workflows to ensure reliable signal detection while preserving key trends in the dataset.

These approaches align with modern data-driven drug discovery frameworks discussed in big data and artificial intelligence in drug discovery.

Algorithm selection

Multiple differential gene expression (DEG) algorithms were evaluated to determine the most suitable statistical approach for the DEL dataset.

After comparative analysis, DESeq2m was selected as the optimal algorithm due to its ability to:

• Handle low-count sequencing data
• Improve statistical reliability
• Detect meaningful enrichment signals

Computational pipeline development

The selected algorithm was integrated into a containerized computational workflow using Nextflow, enabling scalable and reproducible NGS data processing.

The pipeline was deployed on Excelra’s online pipeline platform, ensuring:

• Automated processing of DEL sequencing datasets
• Reproducible analytical workflows
• Scalable computational infrastructure

Our solution

Excelra successfully developed a robust computational pipeline capable of accurately identifying compound hits within large DEL datasets.

Key outcomes

• Development of a hit-calling algorithm optimized for DEL selection data
• Improved detection of true compound hits
• Reduction of false positives caused by noisy sequencing data
• Integration of the algorithm into a scalable Nextflow-based pipeline
• Automated processing of large NGS datasets for high-throughput screening

Key benefits

• Improved hit identification
  The algorithm significantly enhanced detection of promising compounds in DEL screening data.
• Scalable NGS data processing
  The containerized workflow enabled automated and scalable analysis of sequencing datasets.
• Reliable drug discovery insights
  Robust statistical methods improved the accuracy of compound selection for downstream validation.