config.sh.example

#!/usr/bin/env bash
# Paths and pinned resources for the genomic-boundary study.
# Every script sources this file; nothing else hard-codes a path.

# --- source data (read-only) -------------------------------------------------
# Type-material assemblies downloaded with NCBI datasets (see readme in TYPE_DIR).
export TYPE_DIR=/path/to/type_material          # NCBI datasets --from-type download
export GENOME_DIR=$TYPE_DIR/Procariontes          # one directory per accession
export FLAT_DIR=$TYPE_DIR/flat                    # symlinks to every .fna
export META_JSONL=$TYPE_DIR/all_assembly_data_report.jsonl
export ARCHAEA_LIST=$TYPE_DIR/Archaea/ncbi_dataset/Archaea.list

# Pre-existing Mash sketches of the same collection (k=21, seed 42), kept for
# provenance only. The study does not reuse them: they were built by walking the
# file tree and contain four entries that are not genomes (see §2.2). Every
# sketch used here is rebuilt from the canonical manifest into $SKETCH.
export MSH_S1000=$TYPE_DIR/type_30209genomes.msh
export MSH_S10000=$TYPE_DIR/mash10k.msh

# NCBI taxonomy dump, pinned. Downloaded 2026-07-10; the same dump backs the
# RaPDTool benchmark, so labels are comparable between the two studies.
export TAXDUMP=/path/to/taxonkit_dump              # NCBI taxdump (nodes.dmp, names.dmp, …)
export TAXDUMP_DATE=2026-07-10

# --- work area ---------------------------------------------------------------
# Everything the study writes goes here; TYPE_DIR stays read-only.
export WORK=/path/to/work_area                    # needs ~100 GB free
export DATA=$WORK/data
export SKETCH=$WORK/sketch
export DIST=$WORK/dist
export RESULTS=$WORK/results
export FIGURES=$WORK/figures
export LOGS=$WORK/logs
export FUSO=$WORK/fusobacterium   # el caso de Fusobacterium: genomas, sketches y suplementarios
export BINS=$WORK/bins            # bins simulados y su sketch (D13)
export PROTEOME=$WORK/proteome        # proteomas de RefSeq, para el eje de AAI (D7)
export MANUSCRIPT=$WORK/manuscript     # borrador del artículo y su .docx derivado

# --- tools -------------------------------------------------------------------
export BIN=/path/to/conda_env/bin                  # mash, skani, taxonkit
export MASH=$BIN/mash
export SKANI=$BIN/skani
export TAXONKIT=$BIN/taxonkit
export DIAMOND=$BIN/diamond
export DATASETS=${DATASETS:-$HOME/miniconda3/bin/datasets}

export THREADS=${THREADS:-16}

mkdir -p "$FUSO" "$MANUSCRIPT" "$DATA" "$SKETCH" "$DIST" "$RESULTS" "$FIGURES" "$LOGS" "$PROTEOME" "$BINS"

# Python with matplotlib, for the figure step.
export PYVIZ=$BIN/python3