diff --git a/assets/catalog/catalog.md b/assets/catalog/catalog.md
index 452203e..d6e96d0 100644
--- a/assets/catalog/catalog.md
+++ b/assets/catalog/catalog.md
@@ -100,6 +100,12 @@
 - [spliceai/derive](../../modules/nextflow/spliceai/derive/main.nf) — Derive splice event scores from SpliceAI predictions.
 - [spliceai/predict](../../modules/nextflow/spliceai/predict/main.nf) — Predict splice junctions from genomic sequences using SpliceAI.
 - [spliceai/publish](../../modules/nextflow/spliceai/publish/main.nf) — Collect and organize SpliceAI output files.
+- [spliceai2/chunk](../../modules/nextflow/spliceai2/chunk/main.nf) — Chunk genomic sequences for parallel SpliceAI2 prediction.
+- [spliceai2/predict](../../modules/nextflow/spliceai2/predict/main.nf) — Predict splice sites from genomic sequences using SpliceAI2.
+- [spliceai2/publish](../../modules/nextflow/spliceai2/publish/main.nf) — Collect and organize SpliceAI2 output files.
+- [spliceai2/transcripts](../../modules/nextflow/spliceai2/transcripts/main.nf) — Decode transcript structures for a set of loci using SpliceAI2.
+- [subworkflows/spliceai2](../../subworkflows/spliceai2/main.nf) — Chunk, predict, convert, merge, and optionally decode SpliceAI2 outputs.
+- [workflows/spliceai2](../../workflows/spliceai2/main.nf) — Standalone SpliceAI2 pipeline.
 - [aparent/chunk](../../modules/nextflow/aparent/chunk/main.nf) — Chunk genomic regions for parallel APARENT prediction.
 - [aparent/predict](../../modules/nextflow/aparent/predict/main.nf) — Predict polyadenylation sites using the APARENT deep learning model.
 - [minisplice/download](../../modules/nextflow/minisplice/download/main.nf) — Download MiniSplice model and calibration files.
diff --git a/modules/nextflow/spliceai2/chunk/main.nf b/modules/nextflow/spliceai2/chunk/main.nf
new file mode 100644
index 0000000..6ad7a0b
--- /dev/null
+++ b/modules/nextflow/spliceai2/chunk/main.nf
@@ -0,0 +1,59 @@
+/*
+Copyright (c) 2026 The Hiller Lab at the Senckenberg Gessellschaft für Naturforschung
+Distributed under the terms of the Apache License, Version 2.0.
+*/
+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    SPLICEAI2_CHUNK — Chunk genomic sequences for parallel SpliceAI2 prediction.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+process SPLICEAI2_CHUNK {
+    tag "$meta.id"
+    label 'process_low'
+
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        '' :
+        'ghcr.io/hillerlab/spliceai2:latest' }"
+
+    input:
+    tuple val(meta), path(genome)
+    val compression
+
+    output:
+    tuple val(meta), path("chunks/*.fa"),    optional: true, emit: fasta
+    tuple val(meta), path("chunks/*.fa.gz"), optional: true, emit: fasta_gz
+    path "versions.yml",                                  emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def gzip = compression ? '--gzip' : ''
+    """
+    spliceai2 chunk \\
+        $args \\
+        $gzip \\
+        --threads ${task.cpus} \\
+        --sequence "${genome}"
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spliceai2 chunk: \$( spliceai2 chunk --version | sed 's/^spliceai-chunk //g' )
+    END_VERSIONS
+    """
+
+    stub:
+    def extension = compression ? 'fa.gz' : 'fa'
+    """
+    mkdir -p chunks
+    touch chunks/stub.${extension}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spliceai2 chunk: \$( spliceai2 chunk --version | sed 's/^spliceai-chunk //g' )
+    END_VERSIONS
+    """
+}
diff --git a/modules/nextflow/spliceai2/predict/main.nf b/modules/nextflow/spliceai2/predict/main.nf
new file mode 100644
index 0000000..e983b56
--- /dev/null
+++ b/modules/nextflow/spliceai2/predict/main.nf
@@ -0,0 +1,64 @@
+/*
+Copyright (c) 2026 The Hiller Lab at the Senckenberg Gessellschaft für Naturforschung
+Distributed under the terms of the Apache License, Version 2.0.
+*/
+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    SPLICEAI2_PREDICT — Predict splice sites for one sequence chunk.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+process SPLICEAI2_PREDICT {
+    tag "$meta.id"
+    label 'process_low'
+    label 'process_gpu'
+
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        '' :
+        'ghcr.io/hillerlab/spliceai2:latest' }"
+
+    input:
+    tuple val(meta), path(fasta)
+    path models
+    val species
+
+    output:
+    tuple val(meta), path("spliceai/*.donor_plus.wig"),     emit: donor_plus
+    tuple val(meta), path("spliceai/*.donor_minus.wig"),    emit: donor_minus
+    tuple val(meta), path("spliceai/*.acceptor_plus.wig"),  emit: acceptor_plus
+    tuple val(meta), path("spliceai/*.acceptor_minus.wig"), emit: acceptor_minus
+    tuple val(meta), path("spliceai/*.wig"),                emit: all
+    path "versions.yml",                                    emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    spliceai2 predict --no-compile \\
+        $args \\
+        --sequence "${fasta}" \\
+        --models "${models}" \\
+        --species "${species}" \\
+        --outdir spliceai
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spliceai2: \$( spliceai2 --version | sed 's/^spliceai2 //g' )
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    mkdir -p spliceai
+    touch spliceai/stub.donor_plus.wig spliceai/stub.donor_minus.wig \\
+          spliceai/stub.acceptor_plus.wig spliceai/stub.acceptor_minus.wig
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spliceai2: \$( spliceai2 --version | sed 's/^spliceai2 //g' )
+    END_VERSIONS
+    """
+}
diff --git a/modules/nextflow/spliceai2/publish/main.nf b/modules/nextflow/spliceai2/publish/main.nf
new file mode 100644
index 0000000..affe72f
--- /dev/null
+++ b/modules/nextflow/spliceai2/publish/main.nf
@@ -0,0 +1,40 @@
+/*
+Copyright (c) 2026 The Hiller Lab at the Senckenberg Gessellschaft für Naturforschung
+Distributed under the terms of the Apache License, Version 2.0.
+*/
+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    SPLICEAI2_PUBLISH — Collect the four SpliceAI2 strand and site BigWigs.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+process SPLICEAI2_PUBLISH {
+    tag "publish"
+    label 'process_single'
+
+    input:
+    tuple val(meta), path(donor_plus)
+    tuple val(meta1), path(donor_minus)
+    tuple val(meta2), path(acceptor_plus)
+    tuple val(meta3), path(acceptor_minus)
+
+    output:
+    path("spliceai2"), emit: spliceai2
+
+    script:
+    """
+    mkdir -p spliceai2
+    cp "${donor_plus}" spliceai2/
+    cp "${donor_minus}" spliceai2/
+    cp "${acceptor_plus}" spliceai2/
+    cp "${acceptor_minus}" spliceai2/
+    """
+
+    stub:
+    """
+    mkdir -p spliceai2
+    touch spliceai2/donor_plus.bw spliceai2/donor_minus.bw \\
+          spliceai2/acceptor_plus.bw spliceai2/acceptor_minus.bw
+    """
+}
diff --git a/modules/nextflow/spliceai2/transcripts/main.nf b/modules/nextflow/spliceai2/transcripts/main.nf
new file mode 100644
index 0000000..33b7bed
--- /dev/null
+++ b/modules/nextflow/spliceai2/transcripts/main.nf
@@ -0,0 +1,64 @@
+/*
+Copyright (c) 2026 The Hiller Lab at the Senckenberg Gessellschaft für Naturforschung
+Distributed under the terms of the Apache License, Version 2.0.
+*/
+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    SPLICEAI2_TRANSCRIPTS — Decode transcript structures for a set of loci.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+process SPLICEAI2_TRANSCRIPTS {
+    tag "$meta.id"
+    label 'process_high'
+    label 'process_gpu'
+
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        '' :
+        'ghcr.io/hillerlab/spliceai2:latest' }"
+
+    input:
+    tuple val(meta), path(genome), path(genome_fai), path(loci)
+    path models
+    val species
+
+    output:
+    tuple val(meta), path("transcripts/*.transcripts.bed"), emit: transcripts
+    tuple val(meta), path("transcripts/*.skipped.bed"),     emit: skipped
+    path "versions.yml",                                    emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    test -s "${genome_fai}"
+
+    spliceai2 transcripts --no-compile \\
+        $args \\
+        --genome "${genome}" \\
+        --loci "${loci}" \\
+        --models "${models}" \\
+        --species "${species}" \\
+        --outdir transcripts
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spliceai2: \$( spliceai2 --version | sed 's/^spliceai2 //g' )
+    END_VERSIONS
+    """
+
+    stub:
+    def prefix = meta.id
+    """
+    mkdir -p transcripts
+    touch "transcripts/${prefix}.transcripts.bed" "transcripts/${prefix}.skipped.bed"
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        spliceai2: \$( spliceai2 --version | sed 's/^spliceai2 //g' )
+    END_VERSIONS
+    """
+}
diff --git a/modules/nextflow/splicetransformer/derive/main.nf b/modules/nextflow/splicetransformer/derive/main.nf
index 0db9b8d..dbaace5 100644
--- a/modules/nextflow/splicetransformer/derive/main.nf
+++ b/modules/nextflow/splicetransformer/derive/main.nf
@@ -46,7 +46,7 @@ process SPLICETRANSFORMER_DERIVE {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        splicetransformer derive: $( splicetransformer derive --version | sed 's/splicetransformer derive //g' )
+        splicetransformer derive: \$( splicetransformer derive --version | sed 's/splicetransformer derive //g' )
     END_VERSIONS
     """
 
@@ -56,7 +56,7 @@ process SPLICETRANSFORMER_DERIVE {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        splicetransformer derive: $( splicetransformer derive --version | sed 's/splicetransformer derive //g' )
+        splicetransformer derive: \$( splicetransformer derive --version | sed 's/splicetransformer derive //g' )
     END_VERSIONS
     """
 }
diff --git a/subworkflows/spliceai2/main.nf b/subworkflows/spliceai2/main.nf
new file mode 100644
index 0000000..bf181c5
--- /dev/null
+++ b/subworkflows/spliceai2/main.nf
@@ -0,0 +1,122 @@
+/*
+Copyright (c) 2026 The Hiller Lab at the Senckenberg Gessellschaft für Naturforschung
+Distributed under the terms of the Apache License, Version 2.0.
+*/
+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    IMPORT LOCAL MODULES/SUBWORKFLOWS
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+include { SPLICEAI2_CHUNK } from '../../modules/nextflow/spliceai2/chunk/main.nf'
+include { SPLICEAI2_PREDICT } from '../../modules/nextflow/spliceai2/predict/main.nf'
+include { SPLICEAI2_TRANSCRIPTS } from '../../modules/nextflow/spliceai2/transcripts/main.nf'
+include { SPLICEAI2_PUBLISH } from '../../modules/nextflow/spliceai2/publish/main.nf'
+
+include { UCSC_WIGTOBIGWIG as WIGTOBIGWIG_DONOR_PLUS } from '../../modules/nextflow/ucsc/wigtobigwig/main.nf'
+include { UCSC_WIGTOBIGWIG as WIGTOBIGWIG_DONOR_MINUS } from '../../modules/nextflow/ucsc/wigtobigwig/main.nf'
+include { UCSC_WIGTOBIGWIG as WIGTOBIGWIG_ACCEPTOR_PLUS } from '../../modules/nextflow/ucsc/wigtobigwig/main.nf'
+include { UCSC_WIGTOBIGWIG as WIGTOBIGWIG_ACCEPTOR_MINUS } from '../../modules/nextflow/ucsc/wigtobigwig/main.nf'
+
+include { BIGWIGMERGE as BIGWIGMERGE_DONOR_PLUS } from '../../modules/nextflow/bigtools/bigwigmerge/main.nf'
+include { BIGWIGMERGE as BIGWIGMERGE_DONOR_MINUS } from '../../modules/nextflow/bigtools/bigwigmerge/main.nf'
+include { BIGWIGMERGE as BIGWIGMERGE_ACCEPTOR_PLUS } from '../../modules/nextflow/bigtools/bigwigmerge/main.nf'
+include { BIGWIGMERGE as BIGWIGMERGE_ACCEPTOR_MINUS } from '../../modules/nextflow/bigtools/bigwigmerge/main.nf'
+
+workflow SPLICEAI2 {
+    take:
+      genome         // channel: [ val(meta), path(genome) ]
+      chromsizes     // channel: [ path(chromsizes) ]
+      compression    // bool
+      models         // path(model directory)
+      species        // SpliceAI2 species name
+      genome_fai     // path(genome.fai), required with loci
+      loci           // path(BED6), optional
+      ch_versions    // channel: [ path(version) ]
+
+    main:
+      ch_chunks = Channel.empty()
+      ch_transcripts = Channel.empty()
+      ch_skipped = Channel.empty()
+      SPLICEAI2_CHUNK(genome, compression)
+
+      if (compression) {
+        SPLICEAI2_CHUNK.out.fasta_gz
+            .flatMap {
+                meta, fa ->
+                def fas = fa instanceof List ? fa : [fa]
+                fas.collect { it ->
+                    def identifier = it.baseName.replaceFirst(/^tmp\./, '').replaceFirst(/\.fa$/, '')
+                    [ [ id: meta.id + '.' + identifier ], it ]
+                }
+            }
+            .set { ch_chunks }
+      } else {
+        SPLICEAI2_CHUNK.out.fasta
+            .flatMap {
+                meta, fa ->
+                def fas = fa instanceof List ? fa : [fa]
+                fas.collect { it ->
+                    def identifier = it.baseName.replaceFirst(/^tmp\./, '').replaceFirst(/\.fa$/, '')
+                    [ [ id: meta.id + '.' + identifier ], it ]
+                }
+            }
+            .set { ch_chunks }
+      }
+
+      SPLICEAI2_PREDICT(ch_chunks, models, species)
+
+      WIGTOBIGWIG_DONOR_PLUS(SPLICEAI2_PREDICT.out.donor_plus.map { meta, wig -> [ [ id: meta.id + '.donor_plus' ], wig ] }, chromsizes)
+      WIGTOBIGWIG_DONOR_MINUS(SPLICEAI2_PREDICT.out.donor_minus.map { meta, wig -> [ [ id: meta.id + '.donor_minus' ], wig ] }, chromsizes)
+      WIGTOBIGWIG_ACCEPTOR_PLUS(SPLICEAI2_PREDICT.out.acceptor_plus.map { meta, wig -> [ [ id: meta.id + '.acceptor_plus' ], wig ] }, chromsizes)
+      WIGTOBIGWIG_ACCEPTOR_MINUS(SPLICEAI2_PREDICT.out.acceptor_minus.map { meta, wig -> [ [ id: meta.id + '.acceptor_minus' ], wig ] }, chromsizes)
+
+      BIGWIGMERGE_DONOR_PLUS(
+          WIGTOBIGWIG_DONOR_PLUS.out.bigwig
+            .map { meta, bigwig -> bigwig }
+            .collect()
+            .map { bws -> [ [ id : 'donor_plus' ], bws ] },
+      )
+      BIGWIGMERGE_DONOR_MINUS(
+          WIGTOBIGWIG_DONOR_MINUS.out.bigwig
+            .map { meta, bigwig -> bigwig }
+            .collect()
+            .map { bws -> [ [ id : 'donor_minus' ], bws ] },
+      )
+      BIGWIGMERGE_ACCEPTOR_PLUS(
+          WIGTOBIGWIG_ACCEPTOR_PLUS.out.bigwig
+            .map { meta, bigwig -> bigwig }
+            .collect()
+            .map { bws -> [ [ id : 'acceptor_plus' ], bws ] },
+      )
+      BIGWIGMERGE_ACCEPTOR_MINUS(
+          WIGTOBIGWIG_ACCEPTOR_MINUS.out.bigwig
+            .map { meta, bigwig -> bigwig }
+            .collect()
+            .map { bws -> [ [ id : 'acceptor_minus' ], bws ] },
+      )
+
+      SPLICEAI2_PUBLISH(
+          BIGWIGMERGE_DONOR_PLUS.out.bigwig,
+          BIGWIGMERGE_DONOR_MINUS.out.bigwig,
+          BIGWIGMERGE_ACCEPTOR_PLUS.out.bigwig,
+          BIGWIGMERGE_ACCEPTOR_MINUS.out.bigwig,
+      )
+
+      if (loci != null) {
+          SPLICEAI2_TRANSCRIPTS(
+              genome.map { meta, fasta -> [ meta, fasta, genome_fai, loci ] },
+              models,
+              species
+          )
+          ch_transcripts = SPLICEAI2_TRANSCRIPTS.out.transcripts
+          ch_skipped = SPLICEAI2_TRANSCRIPTS.out.skipped
+      }
+
+    emit:
+      spliceai2   = SPLICEAI2_PUBLISH.out.spliceai2
+      transcripts = ch_transcripts
+      skipped     = ch_skipped
+      versions    = ch_versions
+}
diff --git a/workflows/spliceai2/main.nf b/workflows/spliceai2/main.nf
new file mode 100644
index 0000000..ddf70f7
--- /dev/null
+++ b/workflows/spliceai2/main.nf
@@ -0,0 +1,77 @@
+/*
+Copyright (c) 2026 The Hiller Lab at the Senckenberg Gessellschaft für Naturforschung
+Distributed under the terms of the Apache License, Version 2.0.
+*/
+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    IMPORT LOCAL SUBWORKFLOWS
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+include { SPLICEAI2 as RUN } from '../../subworkflows/spliceai2/main.nf'
+include { GENOME } from '../../subworkflows/genome/main.nf'
+
+params.genome              = null
+params.compression         = false
+params.spliceai2_models    = null
+params.spliceai2_species   = 'GRCh38'
+params.spliceai2_loci      = null
+
+workflow SPLICEAI2 {
+    main:
+      if (!params.genome) {
+          error "Missing required --genome"
+      }
+      if (!params.spliceai2_models) {
+          error "Missing required --spliceai2_models"
+      }
+
+      def genome_fasta = file(params.genome, checkIfExists: true)
+      def models = file(params.spliceai2_models, checkIfExists: true)
+      if (!models.toFile().isDirectory()) {
+          error "--spliceai2_models must be a directory"
+      }
+
+      def compression = params.compression?.toString()?.toBoolean() ?: false
+      def species = (params.spliceai2_species ?: 'GRCh38').toString()
+      def genome_fai = null
+      def loci = null
+      if (params.spliceai2_loci) {
+          def uncompressed_genome = genome_fasta.toString().replaceFirst(/\.gz$/, '')
+          def fasta_name = uncompressed_genome.toLowerCase()
+          if (!(fasta_name.endsWith('.fa') || fasta_name.endsWith('.fasta'))) {
+              error "--spliceai2_loci requires a .fa or .fasta genome, optionally compressed as .gz"
+          }
+          genome_fai = file("${uncompressed_genome}.fai", checkIfExists: false)
+          if (!genome_fai.toFile().isFile()) {
+              error "Missing FASTA index '${genome_fai}' required by --spliceai2_loci; index the uncompressed FASTA (e.g. gunzip -k genome.fa.gz && samtools faidx genome.fa) and keep the .fai next to --genome"
+          }
+          loci = file(params.spliceai2_loci, checkIfExists: true)
+          if (!loci.toFile().isFile()) {
+              error "--spliceai2_loci must be a BED file"
+          }
+      }
+
+      GENOME(genome_fasta)
+
+      RUN(
+          GENOME.out.genome.map { g -> [ [ id: g.baseName ], g ] },
+          GENOME.out.chrom_sizes,
+          compression,
+          models,
+          species,
+          genome_fai,
+          loci,
+          GENOME.out.versions
+      )
+
+    emit:
+      spliceai2   = RUN.out.spliceai2
+      transcripts = RUN.out.transcripts
+      skipped     = RUN.out.skipped
+}
+
+workflow {
+    SPLICEAI2()
+}