Carbon A Benchmark Tools#

Configuration File: carbon_a_benchmark_tools.json Tool Type: Local Tools Count: 2

This page contains all tools defined in the carbon_a_benchmark_tools.json configuration file.

Available Tools#

Carbon_A_get_benchmark_genome (Type: CarbonABenchmarkTool)#

Get the files and assembly details of one genome in the Carbon-A evaluation collection. Returns e…

Carbon_A_get_benchmark_genome tool specification

Tool Information:

  • Name: Carbon_A_get_benchmark_genome

  • Type: CarbonABenchmarkTool

  • Description: Get the files and assembly details of one genome in the Carbon-A evaluation collection. Returns every published file with its size, md5 checksum, download URL and role (genomic DNA, NCBI reference annotation, reference CDS, reference proteins, reference RNA, assembly report), plus the per-molecule sequence report: chromosome names, RefSeq accessions, lengths and GC content. Use for: downloading a benchmark genome and its reference annotation, checking an assembly’s size and chromosome count before annotating it, or verifying a download against its checksum. The sequence data itself is not returned, only the URLs to fetch it from.

Parameters:

  • accession (string) (required) NCBI assembly accession of a genome in the collection, for example ‘GCF_000146045.2’ (Saccharomyces cerevisiae). Call Carbon_A_list_benchmark_genomes for the accessions.

  • include_sequence_report ([‘boolean’, ‘null’]) (optional) Include the per-molecule sequence report. Default true.

Example Usage:

query = {
    "name": "Carbon_A_get_benchmark_genome",
    "arguments": {
        "accession": "example_value"
    }
}
result = tu.run(query)

Carbon_A_list_benchmark_genomes (Type: CarbonABenchmarkTool)#

List the eukaryotic genomes in the Carbon-A evaluation collection, the benchmark set published wi…

Carbon_A_list_benchmark_genomes tool specification

Tool Information:

  • Name: Carbon_A_list_benchmark_genomes

  • Type: CarbonABenchmarkTool

  • Description: List the eukaryotic genomes in the Carbon-A evaluation collection, the benchmark set published with the Carbon-A gene-annotation model (HuggingFaceBio/Carbon-A-1.2B). 41 genomes spanning mammals, other vertebrates, invertebrates, plants, fungi and protozoa, each paired with the NCBI reference annotation a prediction is scored against. Returns accession, organism, NCBI taxon id, assembly name, genome size and file count. Filter by the publisher’s split label (‘seen’, ‘unseen’, ‘nonstandard_code’) or by organism name. Use for: choosing a genome to annotate or to benchmark a gene finder against, or finding the reference annotation for one. Sequence payloads are not returned; Carbon_A_get_benchmark_genome gives download URLs and checksums.

Parameters:

  • split ([‘string’, ‘null’]) (optional) Restrict to one eval/ subdirectory. These are the publisher’s own labels, reported as given and not interpreted here: nothing published with the collection defines ‘seen’ and ‘unseen’ further. ‘nonstandard_code’ holds the single genome that uses a non-standard genetic code. Default ‘all’.

  • organism_contains ([‘string’, ‘null’]) (optional) Case-insensitive substring of the organism name, for example ‘Saccharomyces’, ‘sapiens’ or ‘Arabidopsis’.

Example Usage:

query = {
    "name": "Carbon_A_list_benchmark_genomes",
    "arguments": {
    }
}
result = tu.run(query)