Truace
Subscribe

Non-Perturbative Bacterial Identification Directly From Solid Agar Plates Using Raman Micro-Spectroscopy and Machine Learning

Raman spectroscopy is a promising tool for microbial identification, yet its implementation in microbiology and clinical workflow is still restricted due to the accompanying additional preparation required to focus on microbial signals. Here, we demonstrate Raman-based bacterial identification directly from unopened, inverted agar plates, the same conditions used during incubation. Our approach enabled sensitive identification using two Escherichia coli variants, differing only in green fluorescent protein (GFP)…

By the Truvace desk1 min read1 sourcePeer-reviewedTRV-2026-1926· cite & verify
Non-Perturbative Bacterial Identification Directly From Solid Agar Plates Using Raman Micro-Spectroscopy and Machine Learning
Development of nanoporous carbide-derived carbon electrodes for high-performance lithium-ion batteries by Sakamoto, Kamryn M.. Public domain

In brief

Researchers showed that bacteria can be identified by Raman micro-spectroscopy while plates stay closed and inverted, as they are during incubation. The laser path crossed 3-4 mm of background material. A machine learning model was paired with density functional theory calculations to separate the bacterial signal from that background.

This matters because the extra preparation needed to isolate microbial signals has limited Raman use in microbiology and clinical settings. The authors say the method could support real-time monitoring of growth, biofilms and antimicrobial resistance, but those uses are presented as promise rather than demonstrated results.

Main points

  1. The test used two Escherichia coli variants that differed only in green fluorescent protein expression.
  2. Identification worked across diverse media and substrate material conditions.
  3. Classification accuracy exceeded 97.7%, with 10.8% higher mean accuracy and 0.76% less variance than opened-plate measurements.
  4. Raman mapping identified colonies using peaks characteristic of the GFPmut3 chromophore structure generated by DFT.

The gain

Machine learning combined with DFT-based material computation identified E. coli variants through unopened, inverted agar plates at over 97.7% accuracy, avoiding the preparation that has limited Raman use in microbiology workflows.

The rundown

The work targets a practical barrier: Raman spectroscopy is promising for microbial identification, but needs additional preparation to focus on microbial signals. Measuring through an unopened plate removes the step of opening it, matching incubation conditions.

The approach combines traditional DFT-based material computation with machine learning analysis. The authors report that it is robust to changes in algorithms or substrate materials. They also demonstrated colony identification by Raman mapping.

The authors anticipate real-time, non-perturbative monitoring of bacterial growth, biofilm formation and antimicrobial resistance development. These are stated prospects, not outcomes measured in the study.

Sources

  1. Peer-reviewedSmall Methods2026-10-14

The debate