pyproject.toml

[build-system]
requires = ["setuptools", "wheel"]
build-backend = "setuptools.build_meta"

[project]
name = "bit"
version = "1.16.0"
description = "A set of bioinformatics scripts and workflows"
readme = "README.md"
license = { text = "GPLv3" }
authors = [{ name = "Mike Lee", email = "Michael.Lee0517@gmail.com" }]
urls = { Repository = "https://github.com/AstrobioMike/bit" }
# dependencies handled by conda installation

[tool.setuptools]
packages = { find = { where = ["."], include = ["bit*"] } }
include-package-data = true

[tool.setuptools.package-data]
bit = ["tests/data/*", "smk/*.smk", "smk/envs/*"]

[project.scripts]
bit-version = "bit.modules.general:report_version"

## ncbi/GTDB-related ##
# bit-get-accessions-from-GTDB
# get-gtdb-metadata
bit-dl-ncbi-assemblies = "bit.cli.dl_ncbi_assemblies:main"
get-ncbi-assembly-data = "bit.modules.ncbi.get_ncbi_assembly_data:main"
get-ncbi-tax-data = "bit.modules.ncbi.get_ncbi_tax_data:main"
bit-update-ncbi-taxonomy = "bit.modules.ncbi.get_ncbi_tax_data:main"
# bit-get-lineage-from-taxids
# bit-prot-acc-to-taxid

## coverage / mapping ##
bit-cov-analyzer = "bit.cli.cov_analyzer:main"
bit-cov-stats = "bit.cli.cov_stats:main"
bit-get-mapped-reads-pid = "bit.cli.get_mapped_reads_pid:main"

## sequence utilities ##
bit-add-insertion = "bit.cli.add_insertion:main"
bit-calc-gc-per-seq = "bit.cli.calc_gc_per_seq:main"
bit-calc-gc-sliding-window = "bit.cli.calc_gc_sliding_window:main"
bit-calc-variation-in-msa = "bit.cli.calc_variation_in_msa:main"
bit-check-fastq-for-dup-headers = "bit.cli.check_fastq_for_dup_headers:main"
bit-count-bases = "bit.cli.count_bases:main"
bit-dedupe-fasta-headers = "bit.cli.dedupe_fasta_headers:main"
bit-extract-seqs = "bit.cli.extract_seqs:main"
bit-gen-reads = "bit.cli.gen_reads:main"
bit-fasta-to-bed = "bit.cli.fasta_to_bed:main"
bit-fasta-to-genbank = "bit.cli.fasta_to_genbank:main"
bit-filter-fasta-by-length = "bit.cli.filter_fasta_by_length:main"
bit-mutate-seqs = "bit.cli.mutate_seqs:main"
# bit-parse-fastq-by-headers
# bit-remove-wraps
# bit-reorder-fasta
# bit-rename-fasta-headers -> bit-simplify-fasta-headers
# bit-split-multifasta -> move to gist only
bit-summarize-assembly = "bit.cli.summarize_assembly:main"


## table utilities ##
bit-colnames = "bit.cli.colnames:main"
bit-summarize-column = "bit.cli.summarize_column:main"
bit-normalize-table = "bit.cli.normalize_table:main"
# bit-filter-table


## miscellaneous ##
bit-assemble = "bit.cli.assemble:main"
bit-ez-screen = "bit.cli.ez_screen:main"
bit-data-locations = "bit.cli.data_locations:main"
bit-get-test-data = "bit.cli.get_test_data:main"
bit-get-workflow = "bit.cli.get_workflow:main"
bit-lineage-to-tsv = "bit.cli.lineage_to_tsv:main"


## kraken2/bracken helpers ##
bit-kraken2-tax-plots = "bit.cli.kraken2_tax_plots:main"
bit-kraken2-tax-summary = "bit.cli.kraken2_tax_summary:main"
bit-kraken2-combine-tax-summaries = "bit.cli.kraken2_combine_tax_summaries:main"
# bit-combine-bracken-and-add-lineages


## Gene Ontology (GO) helpers ##
# update-go-dbs
# bit-summarize-go-annotations
# bit-slim-down-go-terms
# bit-get-go-term-info

## iToL helpers ##
# bit-gen-iToL-binary-dataset
# bit-gen-iToL-colorstrip
# bit-gen-iToL-map
# bit-gen-iToL-text-dataset
    # probably make a sub-command setup for these if i'm keeping them here
    # (they are probably better suited solely in gtotree)


## genbank manipulations ##
bit-genbank = "bit.cli.genbank:main"


## Database related ##
# get-ncbi-assembly-data = "bit.modules.ncbi.get_ncbi_assembly_data:main"
# get-ncbi-tax-data = "bit.modules.ncbi.get_ncbi_tax_data:main"
# get-go-dbs
# get-gtdb-metadata
    # maybe do a subcommand situation for these
    # bit-data
        # locations
            # check/set
        # get
            # ncbi-assembly-data
            # ncbi-tax-data
            # go-dbs
            # gtdb-metadata