output_formats

Output Formats

Efesto writes results to --out (default: efesto_out/).

---

Standard outputs (always written)

File	Description
Efesto-summary.csv	Per-ORF detailed results with cluster separators; includes cluster_confidence
Efesto-geneSummary-clusters.csv	FeGenie R-script compatible compact summary; includes cluster_confidence
Efesto-heatmap-data.csv	Gene-count matrix (categories × genomes)
Efesto-results-long.tsv	Tidy long-format TSV — one row per ORF, all scoring columns
Efesto-results.gff3	GFF3 feature file (written when --gff_dir or --fna_dir used)
Efesto-summary-stats.tsv	Run-level statistics (counts, confidence tiers, per-genome summaries)

Optional outputs

File	Requires
Efesto-coverage-heatmap.csv	--bam / --bams / --depth / --depths
Efesto-OperonStructure.tsv	--operon_prediction
Efesto-anvio-functions.tsv	--anvio
Efesto-anvio-gene-scores.tsv	--anvio

---

Column reference — Efesto-results-long.tsv

One row per ORF hit. The tidy format for downstream analysis.

Column	Type	Description
category	string	Functional category (e.g. iron_reduction)
genome	string	Source genome filename
contig	string	Source contig identifier
orf	string	ORF identifier (Bakta gene ID when --bakta_gff_dir used)
gene	string	Readable gene name (from FeGenie-map.txt)
hmm_stem	string	HMM file stem
bitscore	float	hmmsearch bitscore
bitscore_cutoff	float	Per-HMM calibrated cutoff (0 = fallback applied)
confidence	string	calibrated or low_confidence
cluster_id	int	Genomic cluster index (unique within genome)
contig_len	int	Contig length in bp
start	int	ORF start coordinate (bp, 1-based)
end	int	ORF end coordinate (bp, 1-based)
strand	string	+ or -
heme_c_motifs	int	Count of CXXCH heme-binding motifs in sequence
cluster_confidence	float	Composite cluster reliability score (0–1.2)
model_nseq	int	Training sequences for this HMM (from registry)
uniop_context	string	UniOP operon ID, singleton_<orf>, or not_in_operon

---

Column reference — Efesto-results.gff3

Standard GFF3 format. Written automatically when coordinate data is available. Loadable directly in IGV, Artemis, and genome browsers.

Fixed columns (GFF3 spec):

Col	Description
seqname	Contig identifier
source	Efesto
feature	gene
start / end	1-based coordinates
score	hmmsearch bitscore
strand	+ or -
frame	.

Attributes (semicolon-separated):

Attribute	Description
ID	ORF identifier
gene	Readable gene name
category	Functional category
hmm_stem	HMM file stem
cluster_id	Cluster index
cluster_confidence	Composite confidence score
confidence	calibrated or low_confidence

---

Column reference — Efesto-summary-stats.tsv

Two-column TSV (section<TAB>key<TAB>value). Sections:

Section	Metrics
RUN	total_orf_hits, total_clusters, genomes_with_hits, genomes_zero_hits, runtime_sec
CONFIDENCE	mean_cluster_confidence, high_confidence_clusters (≥ 0.8), medium_confidence_clusters (0.5–0.8), low_confidence_clusters (< 0.5)
CATEGORY	ORF hit count per functional category (one row per category)
GENOME	Per-genome ORF hit count and mean cluster confidence (one row per genome)

---

Column reference — Efesto-OperonStructure.tsv

Written when --operon_prediction is used.

Column	Description
genome	Source genome
orf	ORF identifier
operon_id	UniOP operon assignment
pair_probability	Co-operon probability with adjacent gene
gene	Readable gene name
category	Functional category

---

Column reference — Efesto-anvio-gene-scores.tsv

Written when --anvio is used. Import with anvi-import-misc-data:

anvi-import-misc-data \
    -c CONTIGS.db \
    --target-data-table genes \
    results/Efesto-anvio-gene-scores.tsv

Column	Description
gene_callers_id	Anvi'o gene caller ID (Bakta ID if --bakta_gff_dir used)
cluster_confidence	Composite cluster reliability score
co_occ_score	Co-occurrence component of confidence
hmm_weight	HMM calibration weight component
uniop_weight	UniOP pair probability component
bgc_boost	BGC boost factor (1.0 or 1.2)

---

Cluster confidence tiers

Score	Tier	Meaning
≥ 0.8	High	Complete operon, calibrated HMMs, compact genomic arrangement
0.5 – 0.8	Medium	Partial operon or low-nseq models involved
< 0.5	Low	Single gene, uncalibrated model, or contig-edge hit

cluster_confidence is capped at 1.2 (possible only when --bgc_dir triggers the 1.2× BGC boost on a near-perfect cluster).

---

Working with outputs in R

FeGenie's original R visualisation script (scripts/plot_heatmap.R) is compatible with Efesto-heatmap-data.csv and Efesto-geneSummary-clusters.csv.

For custom analysis, Efesto-results-long.tsv is the recommended starting point:

library(tidyverse)
hits <- read_tsv("results/Efesto-results-long.tsv")

# High-confidence iron reduction clusters
hits |>
  filter(category == "iron_reduction", cluster_confidence >= 0.8) |>
  group_by(genome, cluster_id) |>
  summarise(genes = paste(gene, collapse=","), confidence = first(cluster_confidence))