Characterization of main strains of ARC inoculant via machine learning-assisted single‐cell Raman spectroscopy

Description of the data and file structure * Ramanome dataset of the ARC inoculant, first edited by Jie Kang on January 16, 2026. File: Ramanome_dataset_of_the_ARC_inoculant_-Kang-_2025.zip Description: Folder Batch 1~4: Generally, for each strain (i.e. Bacillus amyloliquefaciens, Brevibacillus laterosporu, Bacillus mucilaginosus, Enterobacter ludwigii), over 5,000 spectra were collected per batch, with a total of three independent batches measured (i.e. Batch 1~3) to minimize acquisition errors. In total, we constructed a single-cell Raman spectral dataset comprising over 63,000 spectra for the four strains. One more independent validation dataset for the four strains (i.e. Batch 4) was collected using the same methodology to assess model performance. Folder Bacillus amyloliquefaciens (XXXX spectra): Ramanome dataset of the strain Bacillus amyloliquefaciens, which contains XXXX spectra. Folder Brevibacillus laterosporu (XXXX spectra): Ramanome dataset of the strain Brevibacillus laterosporu, which contains XXXX spectra. Folder Bacillus mucilaginosus (XXXX spectra): Ramanome dataset of the strain Bacillus mucilaginosus, which contains XXXX spectra. Folder Enterobacter ludwigii (XXXX spectra): Ramanome dataset of the strain Enterobacter ludwigii, which contains XXXX spectra. For the spectral data files (i.e. the .txt files), the first column represents the wavelength, and the second column represents the spectral intensity at the corresponding wavelength.