Datasets:

neuralbioinfo
/

enterococcus-plasmid-training

pretty_name: Enterococcus clean-v2 plasmid segment dataset
tags:
  - biology
  - genomics
  - plasmid
  - enterococcus
  - prokbert
language:
  - en
configs:
  - config_name: enterococcus_clean_v2_host_assemblies
    data_files:
      - split: train
        path: data/enterococcus_clean_v2_host_assemblies/train-*.parquet
  - config_name: enterococcus_clean_v2_host_contigs
    data_files:
      - split: train
        path: data/enterococcus_clean_v2_host_contigs/train-*.parquet
  - config_name: enterococcus_clean_v2_plasmid_to_assembly_links
    data_files:
      - split: train
        path: data/enterococcus_clean_v2_plasmid_to_assembly_links/train-*.parquet
  - config_name: enterococcus_clean_v2_plasmids
    data_files:
      - split: train
        path: data/enterococcus_clean_v2_plasmids/train-*.parquet
  - config_name: enterococcus_clean_v2_segments
    data_files:
      - split: train
        path: data/enterococcus_clean_v2_segments/train-*.parquet

Enterococcus Clean-v2 Segment Dataset

This dataset contains the Enterococcus clean-v2 NR plasmid training segments and their sampled host candidate segments. Problematic host segments are not dropped from the canonical export; they are retained and tagged.

Configs

config	split	rows	has_sequence
enterococcus_clean_v2_host_assemblies	train	1497	False
enterococcus_clean_v2_host_contigs	train	4530	False
enterococcus_clean_v2_plasmid_to_assembly_links	train	7916	False
enterococcus_clean_v2_plasmids	train	2519	False
enterococcus_clean_v2_segments	train	910900	True

Only enterococcus_clean_v2_segments contains the nucleotide segment sequence column.

Counts

metric	value
plasmid segment rows	182180
host candidate rows	728720
total canonical rows	910900
host rows flagged by plasmid overlap	27417
current clean-training eligible rows	883483
unique source sequence IDs	5133
selected Enterococcus clean-v2 NR plasmids	2519
host assemblies with usable non-plasmid contigs	1497

Segment Flags

segment_id: unique integer primary key, contiguous from 0.
sequence_id: source-sequence grouping ID, contiguous from 0; segments from the same plasmid or host source contig share one value.
labels: 1 for plasmid segments, 0 for host segments.
training_eligible_current_clean: current recommended training filter.
filtered_by_plasmid: host segment has a likely plasmid-overlap minimap2 hit.
is_problematic_segment: currently equivalent to a host segment flagged by plasmid overlap.
problematic_reasons: text reason for problematic rows, currently host_plasmid_overlap.

Loading

from datasets import load_dataset

segments = load_dataset("neuralbioinfo/enterococcus-plasmid-training", "enterococcus_clean_v2_segments", split="train")
clean = segments.filter(lambda row: row["training_eligible_current_clean"])
flagged_hosts = segments.filter(lambda row: row["source_type"] == "host" and row["filtered_by_plasmid"])

Provenance

Source clean-v2 plasmid DB: clean_v2_len2k_4mb_nochrom_broadhq20
Taxonomic scope: Enterococcus genus taxid 1350
Segment length: 4000
Plasmid coverage: 2.0
Host candidate ratio: 4x
Host-overlap filter: minimap2 against selected Enterococcus clean-v2 NR plasmids.

Detailed JSON reports are stored under reports/.