Published by: European Biopharmaceutical Review on 23.10.2024. Link
By Paul Vauterin, Chief Technology Officer at BioLizard
In today’s life sciences industry, economic uncertainties are prompting a shift in investment focus. BioLizard addresses this challenge by empowering biotech companies to unlock the full potential of their data through its unique Bio|Verse framework.
By bridging the gap between complex data and actionable biological insights, the company helps organisations navigate the complexities of modern drug discovery and development (DDD).
The crucial role of data in life sciences development
The journey of drug discovery is inherently long and uncertain. In today’s economic climate, investors are hesitant to finance high-risk early-stage research, favouring later-stage clinical developments instead. This trend puts further pressure on biotech companies to de-risk their pipelines at each stage. They must become very proficient in the collection and assembly of compelling, data-backed evidence that demonstrates a feasible path to success, to convince investors and stakeholders of their projects.
Overcoming the challenges of complex data
Biotech and pharma organisations face significant hurdles due to the sheer volume and complexity of data. They generate vast amounts of proprietary data, while public databases add additional layers of complexity and compatibility issues. The integration of these disparate sources, ensuring accessibility for domain experts, and the translation into coherent insights require an effective engine that turns all this data into actionable insights. While nothing can replace human expertise, tools like Bio|Verse help connect domain experts to the data, and facilitate better decision-making.
Introducing Bio|Verse: tailored data science solutions
The framework combines specialised bioinformatics and data science capabilities with advanced, interactive visual analytics, which empower domain experts to explore and interpret vast amounts of information intuitively. With a focus on user adoption, explainability, and access to the right data at the right time Bio|Verse takes data exploration to the next level.
At BioLizard, we strongly believe that AI models should be transparent and explainable to achieve their true goal. Scientists need to understand what drives certain outcomes, building trust in the results and contributing to a deeper scientific understanding. Therapeutic development is too long and high-risk to rely exclusively on poorly understood black-box solutions for critical steps.
Advancing target and biomarker discovery
One key application within the Bio|Verse framework is Bio|Mx, which facilitates advanced disease-associated target and biomarker discovery through AI applied to multi-omics data – a critical need in modern DDD.
Bio|Mx integrates public and proprietary datasets into a single, coherent picture. This involves complex processes such as data curation, harmonisation, and integration. Graph convolutional neural networks (GCNNs) are used to analyse graph-structured data and leverage biological interactions, to connect data to the ‘map’ of biology – the protein-protein interaction network – and build a comprehensive predictive model.
Deciphering the disease mechanism of prostate cancer
One example is the application of Bio|Mx is in prostate cancer research – a disease with abundant publicly available data but many unresolved questions. By leveraging Bio|Mx, BioLizard addressed key challenges and achieved significant results:
Data integration: We retrieved and merged multi-omics data from seven of the largest public cohorts, including transcriptomics read-outs from biopsies and whole-genome sequencing data
Overcoming challenges: Bio|Mx excelled in seamlessly integrating data from these multiple sources and removing batch effects. This allowed us to complement transcriptomics data from one cohort with genomic information from another
Significant findings: The model predicted 187 genes strongly associated with prostate cancer, many of which were not commonly known to be linked to the disease. These genes represent new options to explore for therapeutic or biomarker targets.
This approach with Bio|Mx not only led to the identification of new, promising genes but also has the potential to help understand the disease mechanism of action – a highly rewarding outcome.
Empowering domain experts through data accessibility
BioLizard places great emphasis on empowering domain experts – biologists, chemists, and biochemists – to interact directly with their data and analysis results. By embracing the philosophy of ‘Everybody can be a data scientist,’ we provide tools that allow specialists to explore and interpret their data without needing advanced technical training. To ensure our solutions work for these specialists, our team of more than 50 experts – the ‘Lizards’ – contribute their expertise in data management, software engineering, bioinformatics, advanced data analytics and AI. Our unique role is to translate data science into something truly meaningful for domain experts in biotech and pharma.
Bio|Verse goes beyond what open-source tools offer by focusing on user adoption, explainability and user experience. It can also leverage large language models (LLMs) to assist interactions between experts and large and complex data sets.
The role of AI in enhancing drug discovery
BioLizard focuses on practical applications of AI that directly benefit drug discovery processes. AI can significantly boost compound screening by expanding the search space. By enhancing screening efficiency and empowering experts to gain insights from vast amounts of data, AI becomes an invaluable tool in DDD.
Additionally, AI tools can help scientists tap into large, unstructured proprietary and public document repositories, ensuring that valuable insights from past research inform current decisions.
A new paradigm for drug discovery
As the life sciences industry continues to evolve, extracting value from complex data becomes increasingly critical. By integrating advanced analytics with user-centric design and explainability at its core, Bio|Verse represents a shift toward more efficient, data- driven decision-making processes in drug discovery.
In this golden age of data collection, the ability to interpret and present data convincingly is crucial. Bio|Verse empowers companies to build the compelling evidence needed to secure investment and advance scientific discovery.