Proteomics has emerged as a groundbreaking tool to fill the gaps that genomics alone cannot. But where are we in the evolution of these methods, and how exactly are they transforming research?
At The Festival of Genomics and Biodata in Boston in June, Benjamin Sun (Director and Head of Population Health Omics, Bristol Myers Squibb), Melissa Miller (Senior Director, Human Genetics, Pfizer) and Sonja Hess (Senior Director, Centre of Genomics Research, Discovery Sciences, Research R&D BioPharmaceuticals, AstraZeneca) came together for an insightful discussion on proteomics at scale and the opportunities ahead.
What are you most excited about at the moment?
One of the most significant breakthroughs highlighted was the scaling up of proteomics, in terms of throughout, analytes and available cohorts. Benjamin Sun highlighted that the maturation of technologies and the availability of resources like those from the UK Biobank has made it possible to analyse tens of thousands of samples, with the number only increasing.
Sonja Hess mentioned the potential of mass-spectrometry (MS) techniques, which can provide deep insights into new biomarkers. These are now being effectively used in small clinical trials, often with no more than 200 patients. This can help researchers identify biomarkers and indicators of prognosis, and efforts are currently focused on increasing the sensitivity and throughput of MS.
What voids can proteomics fill?
In this conversation, genomics was defined as the ‘map’, with proteomics being the path one can take; the proteins are doing all the work, and not necessarily following the blueprint. Although our genomes are static, everyone has different modifications that cannot be ascertained by examining genes in isolation. Proteomics can therefore fill the void of understanding the dynamic nature of biology and disease.
Proteomics also allows for better use of longitudinal studies, which can track changes in biomarkers over time to reveal trends in disease severity and enable early detection. However, these longitudinal samples may not always be available, and proteomics can have more natural and technical variation than genomics. Working at scale can help to minimise and overcome this variation relative to the biological signals being studied.
Current tools can’t capture the whole proteome. Have we reached a limit, and is it important to measure the rest?
The next question posed by Melissa Miller focused on the current limitations of proteomics tools. While platforms like Olink and SomaLogic can measure thousands of proteins, they don’t capture the whole proteome. This is further exacerbated by the presence of protoforms, of which there could be up to a million. However, the speakers agreed that what is currently measurable is still insightful, and progress has been exponential. For example, ten years ago it may have been unbelievable to measure 1,000 proteins, and so perhaps increasing this to 20,00 isn’t so unrealistic. The primary limitation is the availability of purified proteins needed to generate new assays, although progress on this could be quicker than anticipated, especially when complemented by an MS approach.
Another limitation of modern proteomics is that most large-scale studies currently use plasma samples. To fully capture the proteomic, it will be necessary to analyse other tissues matrices, as many proteins are not present in the blood. This presents another challenge for longitudinal studies, as many patients may not be willing to go through invasive tissue biopsies multiple times. Therefore, blood and even urine samples remain highly valuable due to their accessibility.
How can you translate findings from tissue to something found in the blood?
The proteomic changes found in tissues are not always reflected in the blood, and given the aforementioned challenges, it is important to find a way to translate these findings. The speakers suggested a multi-faceted approach, including the use of metabolomics to identify molecules linked to disease progression. They also highlighted imaging as a complementary technique to provide insights that proteomics cannot.
However, a different approach may be required for every different disease. Traditionally, many protocols have been standardised, but there is a need for wider adoption of exploratory approaches, like proteomics and metabolomics, in clinical trials. This would allow for the rediscovery of additional signals and help stratify patients into more homogeneous groups. The ultimate goal is to enable shorter, more cost-effective trials that reach endpoints quicker with bigger effect sizes, and to ensure patients benefit from the right treatments.
What is the next step after the UK Biobank Pharma Proteomics Project?
The conversation wrapped up with discussion about the UK Biobank Pharma Proteomics Project. Although the project will profile over half a million samples, the speakers agreed that a crucial next step is to expand into more diverse cohorts; the UK Biobank dataset is primarily white British, which can lead to predictive models being overfitted to this population. Addressing this diversity issue is a matter of health equity, but will also help to ensure that proteomics is a technique that works for everyone.
