2026 5+6

In an HTC-19 interview, Hans-Gerd Janssen (Unilever) reviews hyphenated chromatography in food analysis. Mass spectrometry and data processing have been the decade's biggest drivers, though accurate peak annotation and integration remain concerns; machine learning, improved spectral databases and automated QC are promising fixes. He notes the difficulty of linking LC–MS peptide data to sensory perception, discusses at-line versus fully integrated systems, and sees sensors, AI and rapid platforms enabling real-time process monitoring.

An overview of how advanced mass spectrometry methods are helping researchers unravel the relationship between food composition and flavour, and of the analytical toolkit now used in modern food and flavour analysis.

SCENT (Spectrum-to-Chemical Embedding alignmeNT) is a multi-modal contrastive learning framework that aligns electron ionisation mass spectra with pretrained chemical structure embeddings, letting models predict odour descriptors from mass spectra alone—no explicit molecular structure needed at inference. It significantly beats MS-only baselines and matches structure-based models, while better approximating continuous human perceptual ratings and generalising to real lab spectra.

An Andy Brunning infographic explaining the chemistry behind smoked foods' classic flavours—how wood-smoke components and reactions deposit and generate the aroma and taste compounds prized in barbecue and smoked products.

Roasted Japanese green tea hojicha is emerging as a flavour trend. Mellow, nutty and low in caffeine with little bitterness, it suits drinks, lattes, desserts, chocolate and even savoury applications. The global market is valued at $2.1 billion, projected to reach $3.9 billion by 2034, with supply expanding partly due to matcha shortages. For industry, hojicha signals a shift toward flavour-led, mainstream-friendly innovation.

Flavour houses are moving beyond amplifying sweetness or savouriness toward engineered sensory experiences using AI, biotechnology and sensory science. Examples include Amano's enzymes that cut bitterness and boost umami, dsm-firmenich's receptor-based off-note masking, Sensient's BioSymphony harmonisers, and Symrise's Symvision AI sensory panels. Driven by demand for healthier, clean-label indulgence, especially in plant-based products, the goal is 'precision indulgence'—craveable flavour that is also sustainable and personalised.

A major review in Addiction casts doubt on alcohol's supposed heart benefits. Genetic (Mendelian randomisation) studies mostly show no protective effect, though the evidence cannot fully rule out a small benefit at low intake. Alcohol is now formally linked to 62 conditions and at least seven cancers, and heavy drinking strongly raises dementia risk. Injury and blood-pressure risks are reversible, but cancer damage is only partly reversible.