Py-GC/MS Dataset for the Classification of Bacterial Species (Version 1.0)

This dataset includes eight bacterial species, which includes biological warfare agent (BWA) simulants and spans both Gram-positive and Gram-negative groups.

Bacillus atrophaeus (DSM 675),

Francisella philomiragia (DSM 7535),

Escherichia coli (DSM 500),

Streptococcus mitis (DSM 12643),

Yersinia enterocolitica (DSM 4780),

Acinetobacter baumannii (DSM 25645),

Agromyces mediolanus (DSM 40), and

Staphylococcus epidermidis (DSM 30905).

Samples were analyzed using pyrolysis gas chromatography mass spectrometry (Py-GC/MS), producing 2D GC×MS chromatograms. From these, 26 fatty acid methyl ester (FAME) features were extracted to capture key biochemical characteristics. The dataset covers three concentration levels, resulting in a total of 104 measurements.
Fatty Acid Methyl Esters (FAMEs) with Retention Time, Carbon Number, and Molecular Weight:

Index
RT (min)
FAME
Carbon Number
MW

1
8.00
Methyl undecanoate
C11:0
200

2
8.06
Methyl 2-hydroxydecanoate
C10:0 2-OH
202

3
8.89
Methyl dodecanoate
C12:0
214

4
9.62
Methyl tridecanoate
C13:0
228

5
9.73
Methyl 2-hydroxydodecanoate
C12:0 2-OH
230

6
10.05
Methyl 3-hydroxydodecanoate
C12:0 3-OH
230

7
10.45
Methyl tetradecanoate
C14:0
242

8
10.89
Methyl 13-methyltetradecanoate
C15:0 iso
256

9
10.97
Methyl 12-methyltetradecanoate
C15:0 anteiso
256

10
11.17
Methyl pentadecanoate
C15:0
256

11
11.18
Methyl 2-hydroxytetradecanoate
C14:0 2-OH
258

12
11.55
Methyl 3-hydroxytetradecanoate
C14:0 3-OH
258

13
11.55
Methyl 14-methylpentadecanoate
C16:0 iso
270

14
11.72
Methyl hexadecenoate (cis-9)
C16:1
268

15*
11.86
Methyl hexadecanoate
C16:0
270

16
12.26
Methyl 15-methylhexadecanoate
C17:0 iso
284

17
12.43
Methyl cis-9,10-methylenehexadecanoate
C17:0
282

18
12.52
Methyl heptadecanoate
C17:0
284

19
12.62
Methyl 2-hydroxyhexadecanoate
C16:0 2-OH
286

20
12.90
Methyl octadecadienoate (all cis-9,12)
C18:2
294

21
13.00
Methyl octadecenoate (cis-9)
C18:1
296

22
13.00
Methyl octadecenoate (trans-9)
C18:2
296

23*
13.15
Methyl octadecanoate
C18:0
298

24
13.64
Methyl cis-9,10-methyleneoctadecanoate
C19:0
310

25
13.70
Methyl nonadecanoate
C19:0
298

26
14.33
Methyl eicosanoate
C20:0
328

*Two FAMEs, Methyl hexadecanoate (C16:0) and Methyl octadecanoate (C18:0), are marked as starred. Their unusually high values, compared to the rest of the features, are attributed to environmental contamination rather than meaningful biological variation. Therefore, these features are not recommended for inclusion in the training process, as they may introduce bias and negatively affect model performance.

The dataset is organized in HDF5 format, with all samples stored under a hierarchical group named samples. Each sample is represented as an individual group (e.g., sample_00000, sample_00001, etc.), ensuring a consistent and scalable structure. 

For every sample, the following data and metadata are included:

2D GCxMS chromatograms (intensity matrix)

m/z axis (mass-to-charge ratio)

retention time axis

FAME features

metadata (original filename, bacterial class, gram type, and concentration level)

The dataset includes the following global information:

Dataset name: Identifier of the dataset.

Description: Overview of the dataset content, including chromatograms, FAME features, bacterial classes, gram types, and concentration levels.

Version: Dataset version.

Source: Origin of the dataset.

DOI: Digital Object Identifier of the dataset.

License: Creative Commons Attribution 4.0 International (CC BY 4.0)

Creators: List of dataset creators.

Contributors: List of dataset contributors.

Contact email: Email address for correspondence and inquiries.

Date: Dataset creation date.

FAME feature names: List of the 26 extracted FAME feature names.