Custom tools import

`samovar tools import`

Register an executable in the install config (tools.<name>) so prepare / exec / resolve_executable find it without a full ./install.sh. Annotator indexes use the same CLI with --type database (writes databases.*, not tools.*).

samovar tools import \
  --name kaiju \
  --env "" \
  --exec kaiju \
  --exec-path /usr/bin/kaiju \
  --type annotator

Flag	Alias	Meaning
`--name`	`-n`	Config key. Also the prepare token (`--kaiju-test "kaiju /db"`).
`--env`		`""` (host) or `conda` (path is a sidecar prefix). Default `""`.
`--exec`		Binary name (default: `--name`). Used as `prefix/bin/<exec>` when `--exec-path` is a conda env directory.
`--exec-path`		File, conda prefix, or a name on `PATH`. If omitted, `PATH` is searched for `--exec` then `--name`.
`--type`	`-t`	Stage / `tool_group`, or `database` / `db` to write `databases.*`.
`--tool`		Annotator that uses this DB (`--type database`). Also accepted as `-n kraken2:dbname`.
`--flags`		Native CLI flags. For a tool: stored on `tools.*`. For a database: merged into that annotator's extra at prepare.
`--lazy-install`		Rebuild recipe for a binary (`@file` or `-` for stdin).
`--lazy-download`		Rebuild recipe for a database (`@file` / `--lazy-download-file`).
`--url`		Official archive URL for a database (feeds `lazy-download`).
`--tool-version`		Version stored in the key (`name:version`).
`--inputs`	`--glob`	Input glob under the run output dir (6th `tools.` slot). Scoring/viz default: `annotations`.

--type values: annotator (a), reads_generator (reads), metagenome_generator (meta), qc (QC, trim), table_reads_generator (table), scoring (score, viz), reprofiler (ml), annotation_converter (converter), workflow, runtime, compiler, database (db).

Databases (`--type database`)

Writes databases.<annotator>.<name:version> (same object shape as tools: path, flags, lazy-download, url, version). Does not go under tools.*.

samovar tools import --type database \
  -n standard_8GB --tool kraken2 \
  --exec-path /path/to/kraken2_db \
  --tool-version 2025oct \
  --flags "--memory-mapping" \
  --url https://genome-idx.s3.amazonaws.com/kraken/k2_standard_08gb_20251015.tar.gz

-n kraken2:standard_8GB is the same as -n standard_8GB --tool kraken2. --exec-path may be a directory, a file (.qza, .fa.gz), or a centrifuge prefix (p_compressed+h+v when .1.cf exists). --lazy-download @fetch.sh stores a custom rebuild script; if omitted, known official URLs get a curl/tar recipe.

Then prepare uses the indexed name, not the filesystem path:

samovar prepare --output_dir RUN --kraken2-test "kraken2 standard_8GB"

flags on the DB record are native tokens for that classifier and are appended to its command. CLI leftovers after the name (--kraken2-test "kraken2 standard_8GB --threads 8") still apply. samovar build --type kaiju --index NAME --flags "…" also registers this object.

samovar export (full) packs the tree plus db-meta/.../lazy-download.sh; unfold install.sh re-imports with --type database.

Tools (`--type annotator` and other groups)

Stored row (legacy list form still parses; on disk the preferred form is a name:version object):

"name": ["env", "workflow", "path", "tool_group"]
"name": ["env", "workflow", "path", "tool_group", "flags"]            # --flags set
"name": ["env", "workflow", "path", "tool_group", "flags", "inputs"]  # scoring, or --inputs set

workflow is conda when --env conda, otherwise bash. path is the resolved file, or the conda prefix. The 5th slot is extra CLI flags from samovar tools import --flags "…". Empty flags are omitted unless a 6th inputs slot is present (scoring/viz).

Those flags are merged at prepare / run with samovar prepare --flags TOOL "…". TOOL can be the imported --name, or the group (annotator / a, table_reads_generator / table). Import flags first, then flags from the prepare token (--myclf-test "myclf /db leftover…"), then --flags.

samovar tools import -n myclf --exec-path /scratch/myclf.py --type annotator \
  --flags "--confidence 0.1"
samovar tools import -n myboil --exec-path /scratch/myboil.py --type table \
  --flags "--log-mu 1 --log-sigma 2"

samovar prepare --output_dir RUN \
  --myclf-test "myclf /path/to/db --threads 8" \
  --flags myclf "--keep-tmp" \
  --table_reads_generator myboil \
  --flags table_reads_generator "--distribution replicates"

--flags annotator "…" applies to every annotator in that prepare. --flags myclf "…" applies only to the run whose type / run_name / command basename is myclf. Same for --flags table_reads_generator vs --flags myboil / --flags camisim-table, --flags reads_generator vs --flags iss / --flags myreads, --flags metagenome_generator vs --flags constant_iss / --flags camisim, --flags qc / --flags trim vs --flags gc_filter, --flags scoring / --flags viz vs --flags counts, and --flags ml / --flags reprofiler vs --flags linear / --flags random_forest.

# Native classifier already on PATH
samovar tools import -n kaiju --exec kaiju --type annotator

# Conda sidecar (prefix, not the file)
samovar tools import -n nanosim --env conda --exec simulator.py \
  --exec-path "$SAMOVAR_ROOT/.cache/samovar/envs/nanosim" --type meta

# Your own annotator script
samovar tools import -n myclf --exec-path /scratch/myclf.py --type annotator
chmod +x /scratch/myclf.py

Then:

samovar prepare --output_dir RUN --myclf-test "myclf /path/to/db"

./install.sh rediscovers PATH tools and keeps existing tools.* values.

How a tool must look (by `--type`)

`annotator`

Input (SamovaR always passes these flags):

Flag	Value
`-i`	R1 FASTQ (`.fastq` or `.fastq.gz`)
`-I`	R2 FASTQ (may be empty; still passed)
`-d`	Database path from prepare (`--name-test "name DB"`)
`-o`	Output file path (create/overwrite)
`-t`	Thread count
extra	Tokens from import `--flags`, leftover after `db` in `--name-test "name DB …"`, and prepare `--flags annotator

Do not require -p unless you are a custom.sh router tool (centrifuge, metauto, assembly_hybrid). Imported annotators are invoked directly. Extra flags are appended at the end of that command (after -t).

Output file (-o): tab-separated, no header, two columns:

<read_id>	<ncbi_taxid>

One row per classified read (skip unclassified / taxid 0 if you want).
read_id is the FASTQ header token without @, without /1 /2.
Empty -o is allowed (zero reads).
Snakemake names the file {sample}_{run}.custom_{name}.out. Downstream parsers only care about the two columns.

Processing: classify reads → write that TSV. Do not write logs into -o. Put temp files next to -o or under $TMPDIR and delete them.

Exit 0 on success even if every read is unclassified (empty file). Non-zero abort the annotators Snakemake rule.

Native names (kraken2, kaiju, metaphlan, …) keep their own CLIs; import only records the binary path. Do not import those under a new --name if you still pass --kaiju-test "kaiju …" — the type is taken from the command basename.

`reads_generator`

Same import pattern as annotators and table regenerators: register a binary, then pass its name to generate and prepare. Extra CLI flags from samovar tools import --flags (5th tools.* slot) are merged with samovar generate --flags / samovar prepare --flags. A reads tool only turns an abundance table (or equal mix from --genome_dir) into FASTQ. Combined community+reads tools use --type meta / --metagenome_generator below.

samovar tools import -n myreads --exec-path /scratch/myreads.py --type reads \
  --flags "--profile hiseq"
chmod +x /scratch/myreads.py

samovar generate --output_dir RUN --genome_dir genomes --host_genome host.fna \
  --reads_generator myreads \
  --flags myreads "--keep-tmp"

samovar prepare --output_dir RUN --reads_generator myreads \
  --flags reads_generator "--threads 4" \
  --kraken2-test "kraken2 /db"

--type reads stores group reads_generator. NanoSim may be --type meta (metagenome_generator); both groups are accepted. --flags reads_generator "…" applies to every reads tool in that run; --flags iss / --flags art / --flags myreads apply only to that name.

Built-ins (no import required):

Name	Generate	Prepare (from abundance tables)
`iss`	InSilicoSeq; FASTQ `{sample}_full_R1/R2.fastq`	Same mixer; `{sample}_{annotator}_R*`
`art` / `illumina`	CAMISIM ART; also `{sample}_illumina_R` then `{sample}_full_R`	ISS mix + `read_type:illumina` in headers and `{sample}_{annotator}_illumina_R*` copies
`wgsim`	CAMISIM wgsim	Same, `read_type:wgsim`
`nanosim` / `ont`	CAMISIM NanoSim3	Same, `read_type:ont`
`hybrid`	CAMISIM one run per tech; headers `read_type:<tech>`	ISS mix (no fake per-tech split)
`camisim`	CAMISIM community design (`--camisim-mode table` by default)	—

--simulator camisim is still valid; --reads_generator art is the same as --simulator camisim --camisim-mode illumina. Import flags on iss / art / wgsim / nanosim are appended to ISS or Nextflow as extra tokens.

Input (custom tools always get these; omit -i when generate uses dirs only):

Flag	Value
`-i`	Abundance CSV (`taxid`, `N_<sample>`…). Optional if `--genome-dir` is set.
`-o`	Output directory (create FASTQ here)
`-m`	Samples metadata CSV (omitted when unset)
`--genome-dir`	Community genomes (generate-style)
`--host-genome` / `--host-fraction`	Host spike-in; fraction or `RANDOM`
`--n-samples` / `--total-reads` / `--seed` / `--model`	Same as `samovar generate`
`--stage`	`generate` or `regenerate`
`--annotator`	Annotator token on prepare (`{sample}_{annotator}_R*`)
`--read-type`	Extra filename id (repeatable: `illumina`, `sequence_type`, …)
`--gzip-reads`	Write `.fastq.gz`
extra	Import `--flags` then prepare/generate `--flags reads_generator\|NAME`

Generate without a table uses --genome_dir, --host_genome, --host_fraction, --n_samples, --total_reads like stock ISS. Generate -i abundance.csv feeds that table instead of a uniform community (ISS and custom tools).

Output FASTQ (.fastq or .fastq.gz):

Headers must keep a recoverable true taxid: taxid:<digits> (ISS combined genomes already do this; CAMISIM harvest adds \|taxid:). If you have a true label under another token, keep taxid: as well so extract_true_taxid works.
Hybrid / multi-tech: add read_type:<token> on the same header. Extra ids may appear in the filename as well as the header: generate {sample}_{sequence_type}_R1.fastq plus the canonical {sample}_full_R1.fastq; prepare {sample}_{annotator}_{sequence_type}_R1.fastq plus {sample}_{annotator}_R1.fastq. Annotators and scoring use the canonical (_full / _{annotator}) stem; tech copies are not classified twice.
Paired outputs: {prefix}_R1.fastq and {prefix}_R2.fastq (R2 may be empty for ONT).

Preferred contract — Python module (--exec-path ends in .py):

from typing import Dict, List, Optional
import pandas as pd

def generate(spec: dict, metadata: Optional[pd.DataFrame], config: dict) -> List[str]:
    # spec has abundance_table, genome_dir, host_genome, host_fraction,
    # n_samples, total_reads, output_dir, stage, annotator, extra_ids, gzip_reads
    # config["extra_argv"] is import + CLI flags
    _ = metadata
    out = []
    # write FASTQ under spec["output_dir"] with taxid: in headers
    return out

Alternatively class ReadsGenerator with generate(self, spec, metadata, config) or run(...). Use if __name__ == "__main__" so import does not run the CLI.

If the name is not in the install config (tools.NAME with group reads_generator or metagenome_generator), generate and prepare error. Import first.

ISS/ART/wgsim/NanoSim native CLIs stay unchanged when you only import them for PATH / conda prefix. Their SamovaR wrappers still tag taxid: / read_type: and extra filename ids as above.

`metagenome_generator`

Combined community design + read simulation in one tool (CAMISIM is the built-in example). Import with --type meta (not --type reads). Used in both samovar generate and samovar prepare.

samovar tools import -n constant_iss \
  --exec-path examples/metagenome_generator/constant_iss.py --type meta \
  --flags "--model hiseq"

samovar generate --output_dir RUN --genome_dir genomes --host_genome host.fna \
  --metagenome_generator constant_iss \
  --flags constant_iss "--n-reads 400"

samovar prepare --output_dir RUN --metagenome_generator constant_iss \
  --flags metagenome_generator "--model hiseq" \
  --test-genomes \
  --kraken2-test "kraken2 /db"

--flags metagenome_generator "…" applies to the combined tool in that run; --flags constant_iss / --flags camisim apply only to that name. Import --flags (5th slot) are merged first.

Built-ins (no import required): camisim (community design + ISS/ART depending on --camisim-mode), hybrid, nanosim. Same as --simulator camisim / --reads_generator hybrid.

Custom contract is the same FASTQ layout as reads_generator (taxid: / read_type:, {sample}_full_R* on generate, {sample}_{annotator}_R* on prepare), but the tool owns abundances: generate may ignore a table and draw a constant/equal community from --genome-dir + host fractions; prepare may flatten observed counts to a constant profile then simulate. Python generate(spec, metadata, config) -> list[str] or the same CLI as reads (-i/-o/--genome-dir/--host-fraction/--stage/--annotator + extra flags).

If NAME is not in the install config with group metagenome_generator, generate and prepare error.

Shipped example: examples/metagenome_generator/constant_iss.py (equal taxid counts + ISS).

`qc`

FASTQ metagenome → trimmed metagenome. Default is absent: setup_reads / ISS output is treated as already trimmed (the qc_initial / qc_generated checkpoints are identity copies into initial_trimmed/ and regenerated_trimmed/). Annotators always read the trimmed dirs.

Import with --type QC (aliases qc, trim, quality). Default is identity. Native short-read trimmers: fastp, Trimmomatic, Cutadapt (Illumina / BGI / MGI postfix). ONT: Chopper, NanoFilt (ont / nanosim3 postfix). A GC-content filter is the Python contract example (tests/tools/gc_filter.py).

# fastp / Trimmomatic / Cutadapt: lazy-install if the binary is not on PATH yet
samovar tools import -n fastp --type QC \
  --exec-path fastp \
  --lazy-install 'conda install -y bioconda::fastp' \
  --pytest

samovar tools import -n trimmomatic --type QC \
  --exec-path trimmomatic \
  --lazy-install 'conda install -y bioconda::trimmomatic' \
  --pytest

samovar tools import -n cutadapt --type QC \
  --exec-path cutadapt \
  --lazy-install 'conda install -y bioconda::cutadapt' \
  --pytest

samovar tools import -n chopper --type QC \
  --exec-path chopper \
  --lazy-install 'conda install -y bioconda::chopper' \
  --pytest

samovar tools import -n nanofilt --type QC \
  --exec-path NanoFilt \
  --lazy-install 'conda install -y bioconda::nanofilt' \
  --pytest

samovar prepare --output_dir RUN --qc fastp \
  --flags qc "--thread 8" \
  --kraken2-test "kraken2 /db"

# hybrid: short-read QC on Illumina, Chopper on ONT
samovar prepare --output_dir RUN --qc fastp \
  --qc-postfix ont:chopper \
  --kraken2-test "kraken2 /db"

samovar tools import -n gc_filter \
  --exec-path tests/tools/gc_filter.py --type QC \
  --flags "--min-gc 0.3 --max-gc 0.7"

--qc fastp (aliases --qc illumina / --qc bgi once fastp is imported), --qc trimmomatic, or --qc cutadapt fills qc_postfix for illumina / bgi / mgi. --qc chopper (aliases --qc ont / --qc nanopore) or --qc nanofilt fills ont / nanosim3. --qc-initial / --qc-generated override one side (none / identity / omit = identity). --flags qc "…" applies to every QC tool; --flags chopper "…" / --flags nanofilt "…" only to that name.

Native CLIs (plus import/prepare extra flags; --threads is translated):

fastp: fastp -i R1 -I R2 -o dest_R1 -O dest_R2 --thread N
Trimmomatic: trimmomatic PE -threads N R1 R2 dest_R1 unpaired1 dest_R2 unpaired2 MINLEN:1 (default step if none given)
Cutadapt: cutadapt --cores N -a … -A … -o dest_R1 -p dest_R2 R1 R2 (dummy poly-G adapter if none given)
Chopper / NanoFilt: stdin FASTQ → stdout (SE; R1 and R2 filtered independently). Defaults -q 0 -l 1 so the contract keeps short reads. Chopper gets --threads N.

Checkpoints: qc_initial (after setup_reads), qc_generated (after sort_reads). YAML keys: qc, qc_initial, qc_generated, qc_postfix: {illumina: fastp, ont: chopper}.

Preferred contract — Python module (--exec-path ends in .py):

from typing import Dict, List, Optional

def trim(r1: str, r2: Optional[str], dest_r1: str, dest_r2: Optional[str], config: dict) -> List[str]:
    # config may include min_gc / max_gc and extra_argv from --flags qc
    # write dest_r1 / dest_r2 (empty files are allowed); keep mates in lockstep
    return [dest_r1] + ([dest_r2] if dest_r2 else [])

Alternatively class QC with trim / run. Use if __name__ == "__main__" so import does not run a CLI.

CLI fallback if there is no trim():

Flag	Value
`-i`	R1 FASTQ
`-I`	R2 FASTQ (optional)
`-o`	Output directory (same filenames as the inputs)
extra	Import `--flags` then prepare `--flags qc\|NAME` (`--min-gc` / `--max-gc` for the GC example)

fastp / Trimmomatic / Cutadapt / Chopper / NanoFilt are native binaries, not the generic -i/-I/-o dir CLI. Import --flags (and --flags-translate, --lazy-install) are appended after SamovaR fills inputs/outputs and threads. Sidecar reports: *.fastp.json, *.cutadapt.json, *.trimlog. samovar tools import --pytest runs tests/test_tool_contracts.py::test_qc_contract. CLI adapters: tests/tools/fastp, cutadapt, trimmomatic, chopper, nanofilt.

`table_reads_generator`

Same idea as a custom annotator: you implement a named tool, import it, then pass that name as the table-regeneration mode.

Built-in modes stay in Python (direct, bootstrap, vae, glm, camisim-table) or optional R (samovar). They all take an Annotation object (the concatenated taxID_* / sample table) and emit per-annotator abundance CSVs (taxid, N_<sample>…). Optional samples metadata is a second table (sample plus any covariates). Built-ins ignore it (null is fine). Your method may use it.

samovar tools import -n myboil --exec-path /scratch/myboil.py --type table
chmod +x /scratch/myboil.py

samovar prepare --output_dir RUN --table_reads_generator myboil \
  --kraken2-test "kraken2 /db"

--regeneration_mode myboil is the same flag. YAML:

regeneration_mode: myboil   # or table_reads_generator: myboil
samples_metadata: /path/to/samples.csv   # optional

If myboil is not in the install config (tools.myboil with group table_reads_generator), prepare and regenerate error. Import first.

Preferred contract — Python module (--exec-path ends in .py):

from typing import Dict, Optional
import pandas as pd
from samovar.parse_annotators import Annotation
from samovar.regenerate import regenerate_preserve

def regenerate(
    annotation: Annotation,
    metadata: Optional[pd.DataFrame],
    config: dict,
) -> Dict[str, pd.DataFrame]:
    # metadata is None unless samples_metadata was set
    _ = metadata
    return regenerate_preserve(
        annotation.DataFrame,
        n_reads=config.get("N_reads"),
    )

config includes N, N_reads, seed, annotation_dir, output_dir, extra_flags / extra_argv (imported --flags plus prepare --flags table_reads_generator "…"), and the rest of the regenerate YAML. Return {annotator: DataFrame} with columns taxid and N_<sample>.

Alternatively define class TableRegenerator with run(self, annotation, metadata, config) (same types).

CLI fallback if there is no regenerate() / TableRegenerator.run:

Flag	Value
`-i`	Annotation directory (`*.annotation.csv`)
`-o`	Output directory (write `{annotator}.csv`)
`-m`	Samples metadata CSV (omitted when unset)
`--seed` / `--N` / `--N_reads`	Same as pipeline config
extra	Tokens from import `--flags` and prepare `--flags table_reads_generator "…"`

Exit 0 after writing abundance CSVs. Use if __name__ == "__main__" so importing the module does not run the CLI.

`scoring`

OPAL and MultiQC stay native: import only records their path. Custom scoring and visualization wrappers use the same --type scoring (score, viz). They run after each viz checkpoint (viz_initial, viz_regenerated, viz_reprofiled) on files already written under the run output directory.

The 6th tools.* slot is the input glob (relative to $out_dir). Built-in viz/OPAL-style tools consume annotation directories samovar_out/*annotations (initial_annotations, regenerated_annotations, reprofiled_annotations). Future methods can point at a combined table, plot dirs, or anything else; store that glob here.

samovar tools import -n counts \
  --exec-path examples/scoring/count_annotations.py --type scoring \
  --inputs '*annotations' \
  --flags "--min-files 0"

samovar tools import -n table_score \
  --exec-path /scratch/table_score.py --type viz \
  --inputs '*annotations/combined_annotation_table.csv'

samovar prepare --output_dir RUN \
  --kraken2-test "kraken2 /db" \
  --flags scoring "--tag extra" \
  --flags counts "--min-files 1"

--scoring NAME (repeatable) runs only those imported tools; omit it to run every custom scoring tool. --scoring none disables them.

`--inputs` value	Matches under `$out_dir`
`*annotations` (default)	`initial_annotations`, `regenerated_annotations`, `reprofiled_annotations`
`annotation_table` / `combined_annotation_table.csv`	`*annotations/combined_annotation_table.csv`
`*_plots` / `plots`	`*_annotations_plots`
comma-separated	union of each pattern

Empty 5th slot is kept when the 6th is set: ["", "bash", "/path/script.py", "scoring", "", "*annotations"]. Dict form: inputs / glob.

Preferred contract — Python module (--exec-path ends in .py):

from pathlib import Path
from typing import List, Optional

def score(inputs: List[Path], output_dir: Path, config: dict) -> None:
    # inputs: resolved paths for this tool's glob
    # output_dir: $out_dir/{name}_scores
    # config["extra_argv"] is import + prepare --flags scoring|NAME
    # config["stage"] is viz_initial / viz_regenerated / viz_reprofiled
    output_dir.mkdir(parents=True, exist_ok=True)
    (output_dir / "ok.txt").write_text("\n".join(str(p) for p in inputs))

Alternatively visualize(...) or class Scorer with score / run. Use if __name__ == "__main__" so import does not run a CLI.

CLI fallback if there is no score():

Flag	Value
`-i`	Each matched path (repeatable)
`-o`	Output directory (`$out_dir/{name}_scores`)
extra	Import `--flags` then prepare `--flags scoring\|NAME`

Native opal / opal.py / multiqc are not re-invoked as custom wrappers (the pipeline already runs compare_annotations, OPAL-adjacent viz, and MultiQC).

Shipped example: examples/scoring/count_annotations.py.

`reprofiler`

Supervised ML on annotation tables with known ground truth (regenerated metagenomes). Built-ins need no import: ensemble (default; RandomForest + AdaBoost, keep the better test score), random_forest / rf, adaboost / ada. Custom methods: --type ml (aliases reprofiler, reprofiling).

The model trains on the regenerated annotation table (true taxid from FASTQ taxid: labels) and may also read ground-truth abundance tables ($out/regenerated/.regenerated_abundance/, the counts used to simulate those reads). It then profiles the initial annotation tables into $out/reprofiled_annotations/ (taxid_SAMOVAR) and writes trained_model.joblib.

samovar tools import -n linear \
  --exec-path examples/reprofiling/linear_classifier.py --type ml \
  --flags "--max-iter 500"

samovar prepare --output_dir RUN --reprofiler linear \
  --flags ml "--C 1" \
  --kraken2-test "kraken2 /db"

--reprofiler / --ml selects the method. --flags ml "…" applies to whichever reprofiler is selected; --flags linear / --flags random_forest apply only to that name.

Preferred contract — Python module:

from typing import Dict
import pandas as pd
from samovar.reprofilers import ReprofileResult, profile_tables_with_model

def reprofile(
    regenerated: pd.DataFrame,
    ground_truth: Dict[str, pd.DataFrame],
    initial: Dict[str, pd.DataFrame],
    config: dict,
) -> ReprofileResult:
    # regenerated: combined regenerated annotation table (column ``true``)
    # ground_truth: {name: abundance CSV} used to generate those metagenomes
    # initial: {sample: annotation table} to profile
    # config["extra_argv"] is import + prepare --flags ml|NAME
    # return tables with taxid_SAMOVAR / taxid_SAMOVAR_confidence, plus model
    ...

CLI fallback: --regenerated, --ground-truth DIR, --initial DIR, -o DIR, plus extra flags. Write {sample}_reprofiled.csv and trained_model.joblib.

See samovar_reprofiling.md.

`workflow` / `runtime` / `compiler`

Snakemake, Nextflow, python, g++: path only.

Built-in annotation tools

centrifuge, metauto, assembly_hybrid, and the dummy constant-taxID classifier still go through src/annotators/custom.sh (-p <name>). Importing centrifuge only updates the binary location used inside that router, not the SamovaR -i/-I/-d/-o contract.

Navigation Menu

Uh oh!

Custom tools import

`samovar tools import`

Databases (`--type database`)

Tools (`--type annotator` and other groups)

How a tool must look (by `--type`)

`annotator`

`reads_generator`

`metagenome_generator`

`qc`

`table_reads_generator`

`scoring`

`reprofiler`

`workflow` / `runtime` / `compiler`

Built-in annotation tools

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Clone this wiki locally

Uh oh!

Custom tools import

samovar tools import

Databases (--type database)

Tools (--type annotator and other groups)

How a tool must look (by --type)

annotator

reads_generator

metagenome_generator

qc

table_reads_generator

scoring

reprofiler

workflow / runtime / compiler

Built-in annotation tools

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Clone this wiki locally

`samovar tools import`

Databases (`--type database`)

Tools (`--type annotator` and other groups)

How a tool must look (by `--type`)

`annotator`

`reads_generator`

`metagenome_generator`

`qc`

`table_reads_generator`

`scoring`

`reprofiler`

`workflow` / `runtime` / `compiler`