Typing
2 of 2 files| File | Description | Rows | Size (.gz) | Download |
|---|---|---|---|---|
| mlst_all.tsv | One row per accession: Pasteur ST, international clone (IC), OXA-51 variant, full allele profile | 48,038 | 356.4 KB | ⬇ Download |
| capsule_all.tsv | Kaptive 3 KL and OCL calls per accession, with country/source/year context | 48,038 | 291.7 KB | ⬇ Download |
Metadata
1 of 1 files| File | Description | Rows | Size (.gz) | Download |
|---|---|---|---|---|
| abyss_metadata.tsv | Country, isolation source, host, collection year, strain, assembly level — one row per accession | 59,160 | 849.1 KB | ⬇ Download |
AMR
2 of 2 files| File | Description | Rows | Size (.gz) | Download |
|---|---|---|---|---|
| amr_all.tsv | One row per (accession × gene): gene, num hits, best identity, class | 440,802 | 1.6 MB | ⬇ Download |
| amr_gene_freq.tsv | One row per gene: num genomes, prevalence %, occurrences | 463 | 3.0 KB | ⬇ Download |
Virulence
2 of 2 files| File | Description | Rows | Size (.gz) | Download |
|---|---|---|---|---|
| vfdb_all.tsv | One row per (accession × gene): gene, num hits, VFC class, VF category | 5,510,501 | 17.3 MB | ⬇ Download |
| vfdb_gene_freq.tsv | One row per gene: num genomes, prevalence %, VFC class | 517 | 3.0 KB | ⬇ Download |
Plasmid
2 of 2 files| File | Description | Rows | Size (.gz) | Download |
|---|---|---|---|---|
| plasmid_all.tsv | One row per (accession × replicon): gene, num hits, best identity | 1,307 | 7.4 KB | ⬇ Download |
| plasmid_gene_freq.tsv | One row per replicon: num genomes, prevalence %, occurrences | 86 | 968 B | ⬇ Download |
📖 How to use these files
Every file is a gzipped TSV. Decompress and open with pandas, R, or the command line:
gunzip mlst_all.tsv.gz head -5 mlst_all.tsv
Or read directly from Python without decompressing — pandas handles .gz automatically:
import pandas as pd
mlst = pd.read_csv("mlst_all.tsv.gz", sep="\t")
caps = pd.read_csv("capsule_all.tsv.gz", sep="\t")
meta = pd.read_csv("abyss_metadata.tsv.gz", sep="\t")
# Join typing + capsule + metadata on accession
df = (mlst.merge(caps, on="accession", how="left")
.merge(meta, on="accession", how="left"))
# IC2 genomes with OXA-23 from bloodstream infections
df[(df["IC"] == "IC2") & (df["source"] == "blood")].head()
🧬 Long-format gene tables
amr_all.tsv, vfdb_all.tsv, and
plasmid_all.tsv are in long format — one row per
(accession, gene) presence. To pivot to a wide presence/absence matrix:
amr_wide = (amr.assign(present=1)
.pivot_table(index="accession", columns="gene",
values="present", fill_value=0))
🔗 GCA ↔ GCF translation
Metadata is GCF-preferred; gene tables are GCA-keyed. All joins
in this dataset use the numeric accession stem, so they work as-is.
If you need to map between NCBI pairs explicitly, use
AB_59160.tsv from the ABYSS repository.
📎 Related resources
- Source genomes: NCBI Assembly — Acinetobacter baumannii
- Download the genomes themselves:
datasets download genome accession --inputfile list.txt - Typing pipeline: AcinetoScope (Scientific Reports 2026)
- Underlying gene data: Shelenkov et al., Data 2024
- Metadata source: NCBI BioSample
📚 Please cite
If you use these data, please cite both source works:
Typing pipeline: Beckley B, Amarh V, Lopes BS, et al. AcinetoScope: a computational pipeline for rapid, comprehensive Acinetobacter baumannii outbreak investigation and resistance gene tracking. Scientific Reports. 2026. doi:10.1038/s41598-026-70189-y
Gene data: Shelenkov A, Mikhaylova Y, Akimkin V. Genomic Epidemiology Dataset for the Important Nosocomial Pathogenic Bacterium Acinetobacter baumannii. Data. 2024;9(2):22. doi:10.3390/data9020022
Please also cite the specific downstream tools you used (mlst, Kaptive, AMRFinderPlus, ABRicate and their reference databases).