The Centre for Molecular Fingerprinting and National Phenome Centre

In this article I will have a look at two different projects of which Prof. Takats is involved with. To understand more about the project, their positives and negatives as well as contribute my own thoughts.

Centre for Molecular Fingerprinting

The Centre for Molecular fingerprinting (CMF) aims to combine modern medicine and generative artificial intelligence to provide a comprehensive picture of the body’s condition by analysing the bloods entire molecular profile [1]. It was founded in 2019 and led by Nobel laureate Ferenc Krausz. It has additional aims to simulate biological time travel allowing researchers to predict the molecular effects of lifestyle changes or aging accurately. But instead of targeting specific variables they are using infrared fingerprinting. So, it does not give the concentration of a particular analyte, but an infrared fingerprint pattern that is generated by essentially the entire repertoire of molecules in the blood. They then compare this pattern with patterns of different phenotypes. They have already seen that the healthy phenotype in fingerprints is very different from lung cancer, prostate cancer, diabetes and other conditions like high blood pressure. Their aim is to create a non-targeted tool used to investigate deviations from the healthy state. They would like to maximise molecular information per sample.

One question that was asked in an interview was. “Can we see in our blood if a new diet or exercise is actually slowing down our aging? This is a very good question. I do not have the answer right now, but this is exactly the direction we would like to go in. So, we have two different research lines or research directions.

One works using real data, where we are trying to see exactly what the effect of these factors you just mentioned — lifestyle factors and so on — is on infrared fingerprints. This is important for understanding which of these factors affect our data.

And the second research line, which again depends on generative modelling, is to essentially try to condition on these lifestyle factors in the same way as we condition on age, sex, and body mass index. This would let us see how a lifestyle change — such as exercise or a medication change — could, in principle, affect the molecular fingerprint of a particular individual.”

They are going for the most cost effective methods but also using other metrological approaches like mass spectrometry proteomics and metabolomics to contribute towards this goal. The completion date is 2027 for both the Hungry 4 Health medical centre and national biobank.

Personal thoughts

I agree with the statement from their paper “… assembling large diverse datasets remains challenging. Collecting repeated visits from the same individuals over extended time periods is particularly demanding, requiring long-term follow-up and substantial logistical coordination. As a result, studies often face limited temporal coverage, imbalanced cohorts, and modest statistical power – constraints that hinder systematic investigation of aging effects, longitudinal health trajectories, and transitions from health to disease.

Generative modelling offers a means to address these limitations by creating synthetic spectra that closely resemble real measurements.”

I agree with the difficulty in sampling over extended periods of time, but I do not see how it accounts for life as it is unexpected and random. Accidents, radiation exposure, diet, air pollution exposure, the environments we are in are changing multiple times every day. Especially the environmental contaminants, if you are tracking an individual the twin must be as close to the person as possible i.e. the greater the time between samples and projection measurement accuracy will be reduced. Additionally, infrared molecular fingerprint looks at wavenumbers between 930 – 3050 cm^-1 how many conditions can that realistically denote, when it is not a discrete variable.

National Phenome Centre

The MRC-NIHR National Phenome Centre was founded in 2012 at Imperial College London. It was created with the fundamental idea that: the genome tells us what might happen, whereas the phenome tells us what is happening. Genetics is relatively stable throughout life, but metabolism continually responds to disease, medication, diet, exercise, environment, the microbiome and ageing.

The phenome centre is a little different in that it uses Nuclear Magnetic Resonance (NMR) spectroscopy and mass spectrometry. They measure metabolites, lipids, lipoproteins and the proteome [2] and use multiple analytical techniques rather than relying on one alone.

Some of the problems from large metabolomics studies are instrument drift, sample preparation differences and batch effects. Instrument drift is the gradual, unwanted change in an instruments output or measurement reading over time, even when the actual measured value stays completely constant. Batch effects are a type of noise or systematic variation in experimental data caused by non-biological factors like different lab technicians, reagent lots, run dates, environmental conditions, instrumental differences… rather than the actual biological conditions being studied. This can lead to false conclusions where sample group together by the day they were processed instead of their true biological state, or when controls were not run on the same day as a sample. Randomisation, meta data tracking and statistical correction are needed to help mitigate these issues.

The phenome centre has a different model in understanding more how metabolism changes during disease. With deep molecular measurement, we can discover biological signatures, identify important molecules, validate them and translate a smaller set of metabolites into clinically usable tests.

Personal Thoughts

Both are extremely useful in their focus on understanding disease but from multiple angles. I do not believe their philosophies are exactly the same, but this is beneficial as you want as many tools as possible in your arsenal to understand and tackle disease. The approach by the CMF is very innovative considering that spectroscopic technology is cheaper, easier to be portable and perhaps very reproducible between instruments. However, they are using mass spectrometry to gain deep insight first to inform the molecular fingerprinting. How well does it work with comorbidities, different populations and ages. Additionally, Raman / IR is good when you know what you have, but the nature of diseases, environment and people is that there will be a lot of variability thus I do not think a cumulative principle component analysis like approach is the best. I think personalised analysis is better.