Complete refactor

.gitignore

@@ -7,8 +7,12 @@ prepUpdate.py
 *.pyc
 include.txt
 build.sh
-update_mtdb
-extract_mtdb
-predb2mtdb
-mtdb
-manage_mtdb
+# legacy root-level console-script copies (anchored so they do NOT clobber the
+# mycotools/mtdb/ package or the test/update_mtdb/ suite of the same name)
+/update_mtdb
+/extract_mtdb
+/predb2mtdb
+/mtdb
+/manage_mtdb
+.gitignore
+CLAUDE.md

Dockerfile

@@ -1,7 +1,6 @@
 FROM mambaorg/micromamba:2.3.0-ubuntu22.04 AS app
 
-ARG MYCOTOOLS_VER="1.0.0"
-USER root
+ARG MYCOTOOLS_VER="2.0.0"
 
 # 'LABEL' instructions tag the image with metadata that might be important to the user
 LABEL base.image="mambaorg/micromamba:2.3.0-ubuntu22.04"
@@ -12,21 +11,28 @@ LABEL description="Mycotools is a compilation of computational biology tools and
 LABEL website="https://github.com/xonq/mycotools"
 LABEL license="https://github.com/xonq/mycotools/blob/master/LICENSE"
 LABEL maintainer="Zachary Konkel"
-LABEL maintainer.email="konkelzach@protonmail.com"
 
 # this is unfortunately necessary to install ete4 
 RUN micromamba install --name base -c conda-forge -c bioconda -c defaults legacy-cgi pip mycotools=${MYCOTOOLS_VER} && \
   eval "$(micromamba shell hook --shell bash)" && \
   micromamba activate base && \
   python3 -m pip install dna_features_viewer && \
-  micromamba clean -a -f -y && \
-  mkdir /data
+  micromamba clean -a -f -y
+
+USER root
+
+RUN mkdir /data
+
+RUN chown $MAMBA_USER:$MAMBA_USER /data
+
+USER MAMBA_USER
 
 ENV PATH="/opt/conda/bin/:${PATH}" \
     LC_ALL=C.UTF-8
 
 # 'CMD' instructions set a default command when the container is run. This is typically 'tool --help.'
 CMD [ "mtdb", "--help" ]
+CMD [ "mycotools", "--help" ]
 
 # 'WORKDIR' sets working directory
-WORKDIR /data
+WORKDIR /data

mycotools/cli.py

@@ -0,0 +1,62 @@
+#! /usr/bin/env python3
+"""Top-level dispatcher for the `mycotools` command.
+
+`mycotools` is the downstream-analysis entrypoint; the database lifecycle
+(update/extract/predb/manage/accession/files) lives under the separate `mtdb`
+command. Routes `mycotools <GROUP> ...` to a group dispatcher, or, for a leaf
+tool like `rename`, directly to that tool's module."""
+from mycotools.lib.subcmd import Dispatcher
+
+# group name/alias -> submodule within the mycotools package. Groups (download,
+# homology, cluster, phylo, stats, seq, gff) are subpackage dispatchers;
+# `rename` is a single leaf module.
+SUBCOMMANDS = {
+    "download": "download",
+    "d": "download",
+    "homology": "homology",
+    "h": "homology",
+    "cluster": "cluster",
+    "c": "cluster",
+    "phylo": "phylo",
+    "p": "phylo",
+    "rename": "rename",
+    "r": "rename",
+    "stats": "stats",
+    "seq": "seq",
+    "gff": "gff",
+}
+
+DESCRIPTION = """Mycotools downstream-analysis toolkit
+
+Groups (all following arguments are forwarded to the group/tool):
+  download   (d)      download genomes/annotations from JGI or NCBI
+  homology   (h)      search query sequence(s) against the database or a fasta
+  cluster    (c)      cluster sequences / circumscribe homology groups
+  phylo      (p)      phylogenies and phylogenetic pipelines (crap/tree/synteny)
+  rename     (r)      substitute MTDB ome codes with taxonomic names in a file
+  stats               annotation / assembly statistics
+  seq                 sequence & coordinate transforms
+  gff                 manipulate / render gff3 files
+
+Database operations (build/manage/query the MTDB) live under the `mtdb` command;
+run `mtdb -h`.
+
+Examples:
+  mycotools download jgi -h
+  mycotools phylo crap -h
+  mycotools homology -h"""
+
+_dispatcher = Dispatcher(
+    "mycotools",
+    "mycotools",
+    SUBCOMMANDS,
+    DESCRIPTION,
+    metavar="GROUP",
+    arg_help="analysis group or tool (see below)",
+)
+main = _dispatcher.main
+cli = _dispatcher.cli
+
+
+if __name__ == "__main__":
+    cli()

mycotools/cluster/__init__.py

@@ -0,0 +1,39 @@
+#! /usr/bin/env python3
+"""Dispatcher for the `mycotools cluster` subcommand.
+
+Routes `mycotools cluster <METHOD> ...` to a clustering module: `db`
+circumscribes database sequences into homology groups, `fasta` runs iterative
+sequence-similarity clustering of a fasta."""
+from mycotools.lib.subcmd import Dispatcher
+
+# subcommand name/alias -> submodule within this package (mycotools.cluster.<module>)
+SUBCOMMANDS = {
+    "db": "db",
+    "fasta": "fasta",
+    "fa": "fasta",
+}
+
+DESCRIPTION = """Group sequences into clusters / homology groups
+
+Methods (all following arguments are forwarded to the method):
+  db              circumscribe database sequences into homology groups
+  fasta   (fa)    iterative sequence-similarity clustering of a fasta
+
+Examples:
+  mycotools cluster db -h
+  mycotools cluster fasta -h"""
+
+_dispatcher = Dispatcher(
+    "mycotools cluster",
+    "mycotools.cluster",
+    SUBCOMMANDS,
+    DESCRIPTION,
+    metavar="METHOD",
+    arg_help="clustering method (see below)",
+)
+main = _dispatcher.main
+cli = _dispatcher.cli
+
+
+if __name__ == "__main__":
+    cli()

mycotools/cluster/__main__.py

@@ -0,0 +1,6 @@
+#! /usr/bin/env python3
+"""Enable ``python -m mycotools.cluster`` to run the cluster dispatcher."""
+from mycotools.cluster import cli
+
+if __name__ == "__main__":
+    cli()