diff --git a/earlgrey_mask.nf b/earlgrey_mask.nf
index d68e1fd..9bc0e25 100644
--- a/earlgrey_mask.nf
+++ b/earlgrey_mask.nf
@@ -56,7 +56,8 @@ def genomeFile(String base) {
 // INCLUDES
 // ════════════════════════════════════════════════════════════════════════════
 
-include { ASM_STATS } from './modules/asm_stats'
+include { ASM_STATS }   from './modules/local/asm_stats'
+include { INPUT_CHECK } from './subworkflows/local/input_check'
 
 // ════════════════════════════════════════════════════════════════════════════
 // PROCESSES
@@ -319,19 +320,15 @@ workflow {
     // without the size/N50 filtering applied by SELECT_REPS.
     def reps
     if (params.skip_select_reps.toBoolean()) {
-        log.info "Skipping SELECT_REPS; processing all genomes for EarlGrey"
-        reps = channel.fromPath(params.samples, glob: false)
-            .splitCsv(header: true)
-            .map { row ->
-                def species = (row.SPECIES?.trim() ?: '')
-                def asmid = (row.ASMID?.trim() ?: '')
-                if (species && asmid) {
-                    "SPECIES,REP_ASMID,REP_SIZE_MB,N_MEMBERS,MEMBER_ASMIDS\n${species},${asmid},0.0,0,"
-                } else {
-                    null
-                }
+        // INPUT_CHECK applies the same taxon/asmid/suppress/n_test filters as funannotate.nf.
+        // We use samples (meta-only, no genome-existence check) because EarlGrey reads genomes
+        // from input_clean_genomes/, not from the raw params.source path.
+        log.info "Skipping SELECT_REPS; processing all filtered genomes for EarlGrey"
+        INPUT_CHECK()
+        reps = INPUT_CHECK.out.samples
+            .map { meta ->
+                "SPECIES,REP_ASMID,REP_SIZE_MB,N_MEMBERS,MEMBER_ASMIDS\n${meta.species},${meta.asmid},0.0,0,"
             }
-            .filter { it != null }
             .collectFile(name: "${launchDir}/misc/repeat_representatives.csv", newLine: false)
     } else {
         reps = SELECT_REPS(
diff --git a/funannotate.nf b/funannotate.nf
index f2a1884..2e6a962 100644
--- a/funannotate.nf
+++ b/funannotate.nf
@@ -1570,137 +1570,6 @@ process FUNANNOTATE_PREDICT {
     """
 }
 
-process ANTISMASH_RUN {
-    label 'antismash'
-    tag "${meta.id}"
-
-    cpus   8
-    memory '16 GB'
-    time   '60h'
-
-    publishDir "${params.target}", mode: 'copy', overwrite: true
-
-    input:
-    val(meta)
-
-    output:
-    tuple val(meta), path("${meta.id}/antismash_local/**")
-
-    script:
-    def out = meta.id
-    def gbk = "${params.target}/${out}/predict_results/${out}.gbk"
-    """
-    # Accept a compressed prediction (.gbk.gz); antismash needs it uncompressed, so
-    # inflate a local copy in the work dir when only the gzipped form is present.
-    GBK="${gbk}"
-    if [ ! -f "\$GBK" ] && [ -f "${gbk}.gz" ]; then
-        zcat "${gbk}.gz" > ${out}.predict.gbk
-        GBK=${out}.predict.gbk
-    fi
-    if [ ! -f "\$GBK" ]; then
-        echo "ERROR: predict GBK not found: ${gbk}[.gz]" >&2
-        exit 1
-    fi
-    source /etc/profile.d/modules.sh 2>/dev/null || true
-    mkdir -p ${out}/antismash_local
-    antismash --taxon ${params.antismash_taxon} \\
-        --output-dir ${out}/antismash_local \\
-        --genefinding-tool none \\
-        --fullhmmer --clusterhmmer --cb-general --pfam2go \\
-        -c ${task.cpus} \\
-        \$GBK
-    pigz ${out}/antismash_local/*.json
-    """
-
-    stub:
-    def out = meta.id
-    """
-    mkdir -p ${out}/antismash_local
-    touch ${out}/antismash_local/${out}.json.gz
-    touch ${out}/antismash_local/index.html
-    """
-}
-
-// IPRSCAN5
-process INTERPROSCAN_RUN {
-    label 'interproscan'
-    tag "${meta.id}"
-
-    cpus   8
-    memory '32 GB'
-    time   '60h'
-
-    publishDir "${params.target}", mode: 'copy', overwrite: true
-
-    input:
-    val(meta)
-
-    output:
-    tuple val(meta), path("${meta.id}/annotate_misc/iprscan.xml")
-
-    script:
-    def out      = meta.id
-    def proteins = "${params.target}/${out}/predict_results/${out}.proteins.fa"
-    """
-    if [ ! -f "${proteins}" ]; then
-        echo "ERROR: protein FASTA not found: ${proteins}" >&2
-        exit 1
-    fi
-    mkdir -p ${out}/annotate_misc
-    interproscan.sh -i ${proteins} -f XML -o ${out}/annotate_misc/iprscan.xml \\
-        -dp -goterms -pa -t p -cpu ${task.cpus}
-    """
-
-    stub:
-    def out = meta.id
-    """
-    mkdir -p ${out}/annotate_misc
-    touch ${out}/annotate_misc/iprscan.xml
-    """
-}
-
-process SIGNALP_RUN {
-    label 'signalp'
-    tag "${meta.id}"
-
-    cpus   8
-    memory '16 GB'
-    time   '12h'
-
-    publishDir "${params.target}", mode: 'copy', overwrite: true
-
-    input:
-    val(meta)
-
-    output:
-    tuple val(meta), path("${meta.id}/annotate_misc/signalp.results.txt")
-
-    script:
-    def out      = meta.id
-    def proteins = "${params.target}/${out}/predict_results/${out}.proteins.fa"
-    """
-    if [ ! -f "${proteins}" ]; then
-        echo "ERROR: protein FASTA not found: ${proteins}" >&2
-        exit 1
-    fi
-    TMPDIR=\${SCRATCH:-/tmp}
-    signalp6 -od \$TMPDIR/${out}_signalp \\
-        -org euk --mode fast -format txt \\
-        -fasta ${proteins} \\
-        --write_procs ${task.cpus} -bs 16
-    mkdir -p ${out}/annotate_misc
-    cp \$TMPDIR/${out}_signalp/prediction_results.txt ${out}/annotate_misc/signalp.results.txt
-    rm -rf \$TMPDIR/${out}_signalp
-    """
-
-    stub:
-    def out = meta.id
-    """
-    mkdir -p ${out}/annotate_misc
-    touch ${out}/annotate_misc/signalp.results.txt
-    """
-}
-
 process FUNANNOTATE_ANNOTATE {
     label 'funannotate'
     tag "${meta.id}"
@@ -1897,7 +1766,11 @@ def staleRnaseq(String out, String species) {
 }
 
 include { validateParameters; paramsSummaryLog; paramsHelp } from 'plugin/nf-schema'
-include { ASM_STATS } from './modules/asm_stats'
+include { ASM_STATS }        from './modules/local/asm_stats'
+include { INPUT_CHECK }      from './subworkflows/local/input_check'
+include { ANTISMASH_RUN }    from './modules/local/antismash_run'
+include { INTERPROSCAN_RUN } from './modules/local/interproscan_run'
+include { SIGNALP_RUN }      from './modules/local/signalp_run'
 
 workflow {
     // `--help` prints schema-driven parameter help (grouped, with types/defaults) and exits.
@@ -1916,74 +1789,13 @@ workflow {
     if( !params.taxondb || !params.funannotate_db )
         error "Missing params.taxondb / params.funannotate_db — add a pipeline profile, e.g. -profile annotate,slurm,module (or use: sbatch nextflow/run_annotate.sh)"
 
-    def suppressSet = (params.suppress && file(params.suppress).exists())
-        ? file(params.suppress).readLines()
-              .collect { it.trim().split(',')[0].trim() }
-              .findAll { it && !it.startsWith('#') }
-              .toSet()
-        : ([] as Set)
-    if (suppressSet) {
-        log.info "Suppress list loaded: ${suppressSet.size()} ASMIDs will be skipped"
-    }
-
-    // ── Taxonomy filter ───────────────────────────────────────────────────────
-    // Parse --taxon RANK:VALUE (e.g. --taxon PHYLUM:Ascomycota).
-    // taxonFilter is a closure applied after splitCsv on the raw row map.
-    def taxonFilter
-    if (params.taxon) {
-        def parts = (params.taxon as String).split(':', 2)
-        if (parts.size() != 2 || !parts[0] || !parts[1]) {
-            error "--taxon must be in RANK:VALUE format, e.g. --taxon PHYLUM:Ascomycota"
-        }
-        def taxRank  = parts[0].toUpperCase()
-        def taxValue = parts[1]
-        log.info "Taxonomy filter: ${taxRank} = '${taxValue}'"
-        taxonFilter = { row -> row[taxRank]?.trim() == taxValue }
-    } else {
-        taxonFilter = { row -> true }
-    }
-
-    // ── ASMID filter ──────────────────────────────────────────────────────────
-    def asmidFilter = params.asmid
-        ? { row -> row.ASMID?.trim() == (params.asmid as String).trim() }
-        : { row -> true }
-    if (params.asmid) {
-        log.info "ASMID filter: processing only '${params.asmid}'"
-    }
-
-    // ── Prediction pipeline ───────────────────────────────────────────────────
-    def jobs = channel.fromPath(params.samples)
-        .splitCsv(header: true)
-        .filter(taxonFilter)
-        .filter(asmidFilter)
-        .map { row ->
-            def meta = SampleUtils.makeMeta(row)
-            def genome_col = row.GENOME?.trim()
-            def gz = genome_col
-                ? (genome_col.startsWith('/') ? file(genome_col) : file("${launchDir}/${genome_col}"))
-                : file("${params.source}/${meta.asmid}/${meta.asmid}_genomic.fna.gz")
-            tuple(meta, gz)
-        }
-        .filter { meta, gz -> meta.id && meta.asmid }
-        .take((params.n_test as int) > 0 ? params.n_test as int : -1)
-        .filter { meta, gz ->
-            if (suppressSet.contains(meta.asmid)) {
-                log.info "Suppressing ${meta.id} (asmid=${meta.asmid})"
-                return false
-            }
-            return true
-        }
-        .filter { meta, gz ->
-            if (!gz.exists()) {
-                log.warn "Missing genome for ${meta.id} (asmid=${meta.asmid}): ${gz}"
-                return false
-            }
-            if (params.debug.toBoolean()) {
-                log.info "Queuing ${meta.id}: genome=${gz} (${gz.size()} bytes)"
-            }
-            return true
-        }
-
+    // ── Samplesheet ingestion (INPUT_CHECK) ──────────────────────────────────
+    // Parses samples CSV, applies taxon/asmid/suppress/n_test filters, builds
+    // meta maps, and resolves genome paths. Two outputs:
+    //   jobs        — tuple(meta, gz)  with genome existence filter (cleaning path)
+    //   postpredict — meta only        no genome filter (annotate/update paths)
+    INPUT_CHECK()
+    def jobs = INPUT_CHECK.out.genomes
     if (params.debug.toBoolean()) {
         jobs.view { meta, gz -> "[CHANNEL] Submitting: out=${meta.id}, asmid=${meta.asmid}, transl_table=${meta.transl_table}, gz=${gz}" }
     }
@@ -2328,14 +2140,9 @@ workflow {
         // postpredict: all samples with a completed predict_results/*.gbk, whether
         // produced in this run or a prior one. This is the source for all optional
         // pre-annotate steps and for FUNANNOTATE_ANNOTATE itself.
-        def postpredict = channel.fromPath(params.samples)
-            .splitCsv(header: true)
-            .filter(taxonFilter)
-            .filter(asmidFilter)
-            .map { row -> SampleUtils.makeMeta(row) }
-            .filter { meta -> meta.id && meta.asmid }
-            .take((params.n_test as int) > 0 ? params.n_test as int : -1)
-            .filter { meta -> !suppressSet.contains(meta.asmid) }
+        // INPUT_CHECK.out.samples already has taxon/asmid/suppress/n_test filters applied;
+        // we just add the predict-results existence check on top.
+        def postpredict = INPUT_CHECK.out.samples
             // Only genomes whose prediction was already complete AND current in a PRIOR run.
             // This is the exact logical complement of the predict_ch filter, so this set is
             // disjoint from the genomes (re)predicted in THIS run (which arrive via
diff --git a/modules/annotation_tools.nf b/modules/annotation_tools.nf
deleted file mode 100644
index 14f0415..0000000
--- a/modules/annotation_tools.nf
+++ /dev/null
@@ -1,145 +0,0 @@
-/*
- * annotation_tools — Post-prediction annotation workflows
- *
- * This module contains optional annotation tools that run after FUNANNOTATE_PREDICT:
- *   - ANTISMASH_RUN: antiSMASH for secondary metabolite detection
- *   - INTERPROSCAN_RUN: InterProScan for protein domain annotation
- *   - SIGNALP_RUN: SignalP for signal peptide prediction
- *
- * These are independent tools that can be run selectively via params:
- *   --run_antismash (default: false)
- *   --run_interpro (default: false)
- *   --run_signalp (default: false)
- *
- * Include in your workflow:
- *   include { ANTISMASH_RUN; INTERPROSCAN_RUN; SIGNALP_RUN } from './modules/annotation_tools'
- */
-
-process ANTISMASH_RUN {
-    label 'antismash'
-    tag "$out"
-
-    cpus   8
-    memory '16 GB'
-    time   '60h'
-
-    publishDir "${params.target}", mode: 'copy', overwrite: true
-
-    input:
-    tuple val(out), val(asmid), val(species), val(strain), val(locustag),
-          val(busco_lineage), val(header_length), val(transl_table)
-
-    output:
-    tuple val(out), path("${out}/antismash_local/**")
-
-    script:
-    def gbk = "${params.target}/${out}/predict_results/${out}.gbk"
-    """
-    # Accept a compressed prediction (.gbk.gz); antismash needs it uncompressed, so
-    # inflate a local copy in the work dir when only the gzipped form is present.
-    GBK="${gbk}"
-    if [ ! -f "\$GBK" ] && [ -f "${gbk}.gz" ]; then
-        zcat "${gbk}.gz" > ${out}.predict.gbk
-        GBK=${out}.predict.gbk
-    fi
-    if [ ! -f "\$GBK" ]; then
-        echo "ERROR: predict GBK not found: ${gbk}[.gz]" >&2
-        exit 1
-    fi
-    source /etc/profile.d/modules.sh 2>/dev/null || true
-    mkdir -p ${out}/antismash_local
-    antismash --taxon ${params.antismash_taxon} \\
-        --output-dir ${out}/antismash_local \\
-        --genefinding-tool none \\
-        --fullhmmer --clusterhmmer --cb-general --pfam2go \\
-        -c ${task.cpus} \\
-        \$GBK
-    pigz ${out}/antismash_local/*.json
-    """
-
-    stub:
-    """
-    mkdir -p ${out}/antismash_local
-    touch ${out}/antismash_local/${out}.json.gz
-    touch ${out}/antismash_local/index.html
-    """
-}
-
-// IPRSCAN5 - InterPro protein domain annotation
-process INTERPROSCAN_RUN {
-    label 'interproscan'
-    tag "$out"
-
-    cpus   8
-    memory '32 GB'
-    time   '60h'
-
-    publishDir "${params.target}", mode: 'copy', overwrite: true
-
-    input:
-    tuple val(out), val(asmid), val(species), val(strain), val(locustag),
-          val(busco_lineage), val(header_length), val(transl_table)
-
-    output:
-    tuple val(out), path("${out}/annotate_misc/iprscan.xml")
-
-    script:
-    def proteins = "${params.target}/${out}/predict_results/${out}.proteins.fa"
-    """
-    if [ ! -f "${proteins}" ]; then
-        echo "ERROR: protein FASTA not found: ${proteins}" >&2
-        exit 1
-    fi
-    mkdir -p ${out}/annotate_misc
-    interproscan.sh -i ${proteins} -f XML -o ${out}/annotate_misc/iprscan.xml \\
-        -dp -goterms -pa -t p -cpu ${task.cpus}
-    """
-
-    stub:
-    """
-    mkdir -p ${out}/annotate_misc
-    touch ${out}/annotate_misc/iprscan.xml
-    """
-}
-
-// SignalP - Signal peptide prediction
-process SIGNALP_RUN {
-    label 'signalp'
-    tag "$out"
-
-    cpus   8
-    memory '16 GB'
-    time   '12h'
-
-    publishDir "${params.target}", mode: 'copy', overwrite: true
-
-    input:
-    tuple val(out), val(asmid), val(species), val(strain), val(locustag),
-          val(busco_lineage), val(header_length), val(transl_table)
-
-    output:
-    tuple val(out), path("${out}/annotate_misc/signalp.results.txt")
-
-    script:
-    def proteins = "${params.target}/${out}/predict_results/${out}.proteins.fa"
-    """
-    if [ ! -f "${proteins}" ]; then
-        echo "ERROR: protein FASTA not found: ${proteins}" >&2
-        exit 1
-    fi
-    TMPDIR=\${SCRATCH:-/tmp}
-    signalp6 -od \$TMPDIR/${out}_signalp \\
-        -org euk --mode fast -format txt \\
-        -fasta ${proteins} \\
-        --write_procs ${task.cpus} -bs 16
-    mkdir -p ${out}/annotate_misc
-    cp \$TMPDIR/${out}_signalp/prediction_results.txt ${out}/annotate_misc/signalp.results.txt
-    rm -rf \$TMPDIR/${out}_signalp
-    """
-
-    stub:
-    """
-    mkdir -p ${out}/annotate_misc
-    touch ${out}/annotate_misc/signalp.results.txt
-    """
-}
diff --git a/modules/local/antismash_run.nf b/modules/local/antismash_run.nf
new file mode 100644
index 0000000..cf14d70
--- /dev/null
+++ b/modules/local/antismash_run.nf
@@ -0,0 +1,50 @@
+process ANTISMASH_RUN {
+    label 'antismash'
+    tag "${meta.id}"
+
+    cpus   8
+    memory '16 GB'
+    time   '60h'
+
+    publishDir "${params.target}", mode: 'copy', overwrite: true
+
+    input:
+    val(meta)
+
+    output:
+    tuple val(meta), path("${meta.id}/antismash_local/**")
+
+    script:
+    def out = meta.id
+    def gbk = "${params.target}/${out}/predict_results/${out}.gbk"
+    """
+    # Accept a compressed prediction (.gbk.gz); antismash needs it uncompressed, so
+    # inflate a local copy in the work dir when only the gzipped form is present.
+    GBK="${gbk}"
+    if [ ! -f "\$GBK" ] && [ -f "${gbk}.gz" ]; then
+        zcat "${gbk}.gz" > ${out}.predict.gbk
+        GBK=${out}.predict.gbk
+    fi
+    if [ ! -f "\$GBK" ]; then
+        echo "ERROR: predict GBK not found: ${gbk}[.gz]" >&2
+        exit 1
+    fi
+    source /etc/profile.d/modules.sh 2>/dev/null || true
+    mkdir -p ${out}/antismash_local
+    antismash --taxon ${params.antismash_taxon} \\
+        --output-dir ${out}/antismash_local \\
+        --genefinding-tool none \\
+        --fullhmmer --clusterhmmer --cb-general --pfam2go \\
+        -c ${task.cpus} \\
+        \$GBK
+    pigz ${out}/antismash_local/*.json
+    """
+
+    stub:
+    def out = meta.id
+    """
+    mkdir -p ${out}/antismash_local
+    touch ${out}/antismash_local/${out}.json.gz
+    touch ${out}/antismash_local/index.html
+    """
+}
diff --git a/modules/asm_stats.nf b/modules/local/asm_stats.nf
similarity index 100%
rename from modules/asm_stats.nf
rename to modules/local/asm_stats.nf
diff --git a/modules/local/interproscan_run.nf b/modules/local/interproscan_run.nf
new file mode 100644
index 0000000..0695da3
--- /dev/null
+++ b/modules/local/interproscan_run.nf
@@ -0,0 +1,37 @@
+// IPRSCAN5
+process INTERPROSCAN_RUN {
+    label 'interproscan'
+    tag "${meta.id}"
+
+    cpus   8
+    memory '32 GB'
+    time   '60h'
+
+    publishDir "${params.target}", mode: 'copy', overwrite: true
+
+    input:
+    val(meta)
+
+    output:
+    tuple val(meta), path("${meta.id}/annotate_misc/iprscan.xml")
+
+    script:
+    def out      = meta.id
+    def proteins = "${params.target}/${out}/predict_results/${out}.proteins.fa"
+    """
+    if [ ! -f "${proteins}" ]; then
+        echo "ERROR: protein FASTA not found: ${proteins}" >&2
+        exit 1
+    fi
+    mkdir -p ${out}/annotate_misc
+    interproscan.sh -i ${proteins} -f XML -o ${out}/annotate_misc/iprscan.xml \\
+        -dp -goterms -pa -t p -cpu ${task.cpus}
+    """
+
+    stub:
+    def out = meta.id
+    """
+    mkdir -p ${out}/annotate_misc
+    touch ${out}/annotate_misc/iprscan.xml
+    """
+}
diff --git a/modules/local/signalp_run.nf b/modules/local/signalp_run.nf
new file mode 100644
index 0000000..6b03d85
--- /dev/null
+++ b/modules/local/signalp_run.nf
@@ -0,0 +1,41 @@
+process SIGNALP_RUN {
+    label 'signalp'
+    tag "${meta.id}"
+
+    cpus   8
+    memory '16 GB'
+    time   '12h'
+
+    publishDir "${params.target}", mode: 'copy', overwrite: true
+
+    input:
+    val(meta)
+
+    output:
+    tuple val(meta), path("${meta.id}/annotate_misc/signalp.results.txt")
+
+    script:
+    def out      = meta.id
+    def proteins = "${params.target}/${out}/predict_results/${out}.proteins.fa"
+    """
+    if [ ! -f "${proteins}" ]; then
+        echo "ERROR: protein FASTA not found: ${proteins}" >&2
+        exit 1
+    fi
+    TMPDIR=\${SCRATCH:-/tmp}
+    signalp6 -od \$TMPDIR/${out}_signalp \\
+        -org euk --mode fast -format txt \\
+        -fasta ${proteins} \\
+        --write_procs ${task.cpus} -bs 16
+    mkdir -p ${out}/annotate_misc
+    cp \$TMPDIR/${out}_signalp/prediction_results.txt ${out}/annotate_misc/signalp.results.txt
+    rm -rf \$TMPDIR/${out}_signalp
+    """
+
+    stub:
+    def out = meta.id
+    """
+    mkdir -p ${out}/annotate_misc
+    touch ${out}/annotate_misc/signalp.results.txt
+    """
+}
diff --git a/subworkflows/local/input_check.nf b/subworkflows/local/input_check.nf
new file mode 100644
index 0000000..6dfd251
--- /dev/null
+++ b/subworkflows/local/input_check.nf
@@ -0,0 +1,104 @@
+/*
+ * INPUT_CHECK — canonical samplesheet ingestion
+ *
+ * Parses params.samples (CSV with header), applies taxon / ASMID / suppress /
+ * n_test filters, builds SampleUtils.makeMeta(row), and resolves the genome
+ * file from the GENOME column or the NCBI_ASM source directory.
+ *
+ * Emits two channels so callers can choose the right level of filtering:
+ *
+ *   samples  — val(meta)              filtered by taxon/asmid/suppress/n_test
+ *                                     but NOT by genome existence (use for any
+ *                                     workflow that just needs sample metadata,
+ *                                     e.g. postpredict or earlgrey select path)
+ *
+ *   genomes  — tuple(val(meta), path(gz))
+ *                                     additionally filtered to samples whose
+ *                                     raw genome file exists on disk (use as
+ *                                     the primary jobs channel in funannotate)
+ *
+ * All filtering logic lives here so it does not have to be duplicated across
+ * funannotate.nf, earlgrey_mask.nf, or any future entrypoint.
+ */
+
+workflow INPUT_CHECK {
+
+    main:
+
+    // ── Suppress list ────────────────────────────────────────────────────────
+    def suppressSet = (params.suppress && file(params.suppress, glob: false).exists())
+        ? file(params.suppress, glob: false).readLines()
+              .collect { it.trim().split(',')[0].trim() }
+              .findAll { it && !it.startsWith('#') }
+              .toSet()
+        : ([] as Set)
+    if (suppressSet) {
+        log.info "Suppress list loaded: ${suppressSet.size()} ASMIDs will be skipped"
+    }
+
+    // ── Taxonomy filter ──────────────────────────────────────────────────────
+    // Parse --taxon RANK:VALUE (e.g. --taxon PHYLUM:Ascomycota).
+    def taxonFilter
+    if (params.taxon) {
+        def parts = (params.taxon as String).split(':', 2)
+        if (parts.size() != 2 || !parts[0] || !parts[1]) {
+            error "--taxon must be in RANK:VALUE format, e.g. --taxon PHYLUM:Ascomycota"
+        }
+        def taxRank  = parts[0].toUpperCase()
+        def taxValue = parts[1]
+        log.info "Taxonomy filter: ${taxRank} = '${taxValue}'"
+        taxonFilter = { row -> row[taxRank]?.trim() == taxValue }
+    } else {
+        taxonFilter = { row -> true }
+    }
+
+    // ── ASMID filter ─────────────────────────────────────────────────────────
+    def asmidFilter = params.asmid
+        ? { row -> row.ASMID?.trim() == (params.asmid as String).trim() }
+        : { row -> true }
+    if (params.asmid) {
+        log.info "ASMID filter: processing only '${params.asmid}'"
+    }
+
+    // ── Build channel with meta + resolved genome path ───────────────────────
+    // ch_raw passes taxon/asmid/suppress/n_test filters; gz resolution is
+    // done here so both emit channels share the same genome-path logic.
+    def ch_raw = channel.fromPath(params.samples)
+        .splitCsv(header: true)
+        .filter(taxonFilter)
+        .filter(asmidFilter)
+        .map { row ->
+            def meta       = SampleUtils.makeMeta(row)
+            def genome_col = row.GENOME?.trim()
+            def gz = genome_col
+                ? (genome_col.startsWith('/') ? file(genome_col) : file("${launchDir}/${genome_col}"))
+                : file("${params.source}/${meta.asmid}/${meta.asmid}_genomic.fna.gz")
+            tuple(meta, gz)
+        }
+        .filter { meta, gz -> meta.id && meta.asmid }
+        .take((params.n_test as int) > 0 ? params.n_test as int : -1)
+        .filter { meta, gz ->
+            if (suppressSet.contains(meta.asmid)) {
+                log.info "Suppressing ${meta.id} (asmid=${meta.asmid})"
+                return false
+            }
+            return true
+        }
+
+    // ── genomes: additionally require the genome file to exist ────────────────
+    def ch_genomes = ch_raw
+        .filter { meta, gz ->
+            if (!gz.exists()) {
+                log.warn "Missing genome for ${meta.id} (asmid=${meta.asmid}): ${gz}"
+                return false
+            }
+            if (params.debug.toBoolean()) {
+                log.info "Queuing ${meta.id}: genome=${gz} (${gz.size()} bytes)"
+            }
+            return true
+        }
+
+    emit:
+    samples = ch_raw.map { meta, _gz -> meta }
+    genomes = ch_genomes
+}
diff --git a/workflows/.gitkeep b/workflows/.gitkeep
new file mode 100644
index 0000000..e69de29