3 Ways Machine Learning Propels Metabolomics at Metabolon

At Metabolon, we have spent the last 20 years building the best metabolomic competency in the world. Today, we enable, accelerate, and support drug development through biomarker discovery, understanding mechanism of action, patient stratification and more. One tool we use to power our work is machine learning, which automates routine tasks and teachable processes so our experts can focus their efforts on the challenges that most require their expertise.

Metabolon measures biochemical changes through the study of small molecules to provide a profound understanding of biological systems. Data science enables the continuous improvement of our industry-leading capabilities, allowing us to provide a broader view beyond any other metabolomic provider. As part of our data science initiatives, we leverage machine learning to extract insight and recognize data patterns computationally. Metabolon’s machine learning capability improves our data curation throughput, identifies biochemicals’ signatures, and detects anomalies to accurately and rapidly ensure quality control. Our data science and machine learning capabilities provide improved coverage, quality, turnaround time, and interpretations and, ultimately, study success for our clients.

Machine learning’s effectiveness is strongly correlated to the accuracy, breadth, and consistency of the data on which it is trained. Long before the rise of data science, Metabolon professionals were generating expertly curated data sets, precise statistical analysis, and reports containing deep biological insights. Here are three examples of machine learning at work at Metabolon:

1. Streamlined Quality Control

In the lab, time is always of the essence and attention to detail is paramount. Late-stage analysis revealing flaws in original data can invalidate days or weeks of work, de-railing timelines and production deliverables. Machine learning enables us to provide clients a compressed quality control cycle to detect failure modes much earlier than is possible with a purely manual process.

For example, with several product lines, stringent requirements are tested via statistical analysis once curation is complete. A fully manual process would require several days of work. The auto curation utility combined with automated statistical analyses targeted at product requirements has enabled the classification of those samples that would likely fail quality control checks days later. Back-up samples may then immediately run, decreasing turnaround time for the client and reducing the amount of labor spent on samples doomed to fail at a downstream quality control step.

Metabolon has demonstrated the utility of machine learning approaches that provide results approximating expert-driven, post-processing. Detecting failure modes instantly after initial raw data production and prompting targeted human review results in significant time savings, more rapid average product delivery, and reduced labor, which can instead focus on shippable results rather than reruns.

2. Faster Data Curation

In textbook machine learning, labeling problems are solved by learning a classification model from unbiased ground truth data. In real applications, however, the matter may be significantly complicated by the practices and protocols used to produce the training data. As an example, Metabolon generates LC/MS mass spectroscopy data from which metabolites’ presence is inferred. Historically, expert curators diligently examined these data with software assistance to confirm or deny the compounds’ presence. Each sample processed on our platforms is examined for the presence or absence of every one of the 5200+ tier one identified compounds in Metabolon’s metabolite knowledgebase.

Machine learning allows us to achieve this same high-quality data, but much faster. We can bring a data set directly to quality control through machine learning, saving time by automatically performing initial curation. Machine learning also allows us to quickly determine with certainty which compounds are present and the ones that were never present, significantly reducing or even eliminating the need for human experts to make these trivial decisions.

Built on historical curations, machine learning feeds an auto curation utility that can curate many routine compounds. Consider cholesterol, which is readily found in human plasma and, therefore, not an efficient use of staff expertise. With the support of machine learning tools, we can leverage our human expert curators’ skills in delving for the presence of ethylparaben sulfate, which often presents with interfering ions, or differentiating between compounds like isoleucylglycine and alanylvaline. These compounds are not chromatographically separated but have distinct MS/MS fragmentation, usually containing a tremendous variety of information.

3. Continued Knowledge Expansion

At Metabolon we’re proud of our metabolite knowledgebase and industry-leading expertise, but we’re not stopping there. There is still much to be discovered in terms of new compounds and their impact on life sciences research and drug development, and we’re committed to unlocking that knowledge to deliver game-changing insights. The body of literature, reports, and insights produced by Metabolon’s internal experts have served as the basis for a collaboration with data science to develop a shared vocabulary, a knowledgebase, and software to support the recording of continued knowledge expansion. Machine Learning enables our experts to leverage our vast internal experience of highly curated pathway knowledge, disease profiles, and the world’s best metabolomic insights. Future data science collaborations with these experts will automatically surface relevant historical knowledge to expert staff on each and every experiment run with Metabolon.

At Metabolon, machine learning is a crucial ingredient to our clients’ success. Our machine learning approaches have resulted in tremendous time savings, more rapid average product delivery, and labor, which is more often spent on shippable results than reruns. We’re putting these efforts to work for clients every day to produce actionable insights and introduce new therapies faster. Want to learn more about our process and how a machine-learning enhanced metabolomics study can further your research? Contact us today at [email protected] to get started.