Output files

Which file gifter reads, and which files preserve the evidence and run context behind it.

Annotation output

A completed giftag annotate run writes three files in the requested output directory and, for nucleotide input, a proteins/ directory. The tables are tab-separated with a header row.

giftag_markers.tsv

One row per gene and accepted gifter marker. A gene with more than one accepted marker has more than one row. gifter reads the first four fields; the remaining fields let you inspect the search evidence.

Column	Meaning
genome_id	Identifier derived from the input file name.
gene_id	Protein or predicted-gene identifier.
namespace, accession	The namespaced marker identity in gifter’s catalogue.
source, profile	Profile collection and specific HMM supporting the marker.
rule	Source-specific acceptance rule applied to the hit.
score, evalue, threshold	Values behind the accepted hit; their interpretation follows rule.
coverage	HMM coverage when the source’s rule supplies it.
target_from, target_to	Hit coordinates on the target protein when available.

Rows are ordered by genome, then by gene and marker within each genome. If more than one hit supports the same gene and marker, giftag retains the strongest one. A dbCAN hit to an official subfamily model such as GH43_18 also licenses its broader GH43 family marker; the broader row retains GH43_18 in profile so the evidence remains visible. The reverse implication does not hold.

giftag_genomes.tsv

One row per input genome, with genome_id, input, input_type, gene_calling, sequences, length_bp, proteins, marker_genes and markers. It records what was processed and how many distinct genes and marker identities were found. length_bp is empty for protein FASTA.

giftag_run.json

The run manifest records giftag’s version and finish time, the built database path and gifter database identity, source releases, marker-status counts, search parameters, and the number of genomes and marker rows. giftag writes it last. If the run stops partway through, completed genomes remain in the TSV files but this JSON file is absent; do not present those files as a completed run.

proteins/<genome_id>.faa

Predicted proteins from nucleotide input. For protein FASTA input, giftag searches the supplied sequences and writes no predicted-protein file.

Built database files

The directory produced by giftag build is separate from annotation output:

File	Purpose
giftag.json	Database format and build time; giftag and gifter versions; source releases, locations, checksums and terms; marker-status counts.
markers.tsv	Every marker requested by gifter and whether giftag could search it. See Marker coverage.
profiles.tsv	Searchable profiles with their rule and threshold.
Source .hmm files	The local profile libraries giftag uses for annotation.

Keep the database manifest and giftag_run.json with a published analysis. The marker TSV alone says which hits were accepted, but not which profiles were available to be searched.