Skip to contents
Chapters and reference

Complete canonical API text; task articles explain the workflows. Usage guide

Reads local signature TSV or cached entry JSON from InterPro/Pfam without running an annotation engine or retrieving online data. Accession case and versions remain verbatim, including lowercase accessions returned by the API. Declared full protein lengths supply seq_data; feature endpoints never stand in for full lengths. No source characters or verification are invented.

Usage

R · Example code
read_signature_matches(
  file = NULL,
  files = NULL,
  format = c("scan_tsv", "entry_json"),
  source_file = FALSE
)

Arguments

file

Optional local path vector or glob patterns, before files.

files

Optional additional input paths or patterns.

format

Explicit schema: "scan_tsv" (default, 11-15 columns) or "entry_json" (cached entry results with metadata/proteins/locations). JSON requires optional jsonlite. No pagination links are followed; declared counts and remaining-page links stay in the report.

source_file

Add .source_file to feature rows, default FALSE.

Value

A list with seq_data (aa), features$signature and report. Source lengths, native metadata, raw files and pagination remain traceable. Known length or declared MD5 conflicts for the same accession fail, rather than silently merging different sources. A declared MD5 is retained, not independently verified without actual source characters.

Details

Source score is a list-column preserving each native scalar/string/missing value. Member-library scores and e-values cannot be assumed comparable; no universal confidence, evalue, Identity or read support is manufactured. Original TSV fields and JSON objects remain in the report. TSV field 10, dates, optional integrated/GO/pathway metadata are retained without guessing version-specific semantics or promoting them to plotting instructions.

Each JSON location remains one feature row, reusing the existing segments contract for actual fragments and their original indices/status. start/end bound that location; gaps between its segments are not filled. Per-file .source_match and entry/protein/location indices retain the source context. Model and signature IDs are labels, not unique feature-instance IDs; feature_id stays missing. All representative and nonrepresentative matches remain. TSV rows retain their original interval and physical line. No grouping across TSV rows is guessed. Match indices need .source_input when several files are combined; none is an invented biological observation ID.

Examples

R · Example code
# Explicit re-encoding of cached AA annotations, not a new analysis run.
path <- system.file("extdata", "P00517-signatures.tsv", package = "ggchord")
matches <- read_signature_matches(path)
matches$features$signature[c("accver", "start", "end", "signature_id")]
#>   accver start end signature_id
#> 1 p00517    44 298      PF00069

Search documentation

↑ ↓ select · Enter open · Esc close; or browse by task. API and functions