Simulated Metagenomic Reads for Mycoplasma Bovis Strain Classification

Data Sheet 1 presents in silico performance results for a targeted enrichment sequencing approach to classify Mycoplasma bovis strains in milk. The dataset is based on 620 M. bovis whole-genome sequences from NCBI, with 162 originating from milk, and simulates mock milk samples with varying strain mixtures. It was authored by Marit M. Biesheuvel and last updated on March 18, 2026.

Use Cases

• Benchmarking classification tools like Kraken2 and Themisto/mSWEEP based on simulated metagenomic reads.
• Evaluating the effect of enrichment proportion and strain mixture complexity on classification accuracy.
• Assessing the suitability of Genomically Clustered Sequence Variants (GSVs) for strain-level microbial epidemiology.

Strengths

• Performance metrics are provided for two classification tools, including specific accuracy (84.9% vs 1.4%), sensitivity, and PPV values.
• Simulation parameters are clearly defined, including 1 to 9 GSVs per sample and enrichment proportions from 0.3 to 0.9.
• The underlying data is sourced from 620 publicly available M. bovis whole-genome sequences from NCBI.

Limitations

• The dataset is a 237.2 KB PDF; the underlying simulated read data is not directly provided, limiting computational reuse.
• The data is entirely synthetic, generated in silico from existing genomes; its biological relevance requires further validation.
• Row count and column-level documentation for any underlying tabular results are unknown.

Provenance

Source

Marit M. Biesheuvel via figshare

Collection Method

In silico simulation using 620 M. bovis whole-genome sequences downloaded from NCBI, fragmented into 150-bp reads.

Freshness

Last updated 2026-03-18 15:43:11