From 73b4fd3b8afd102a6b257b7c4c5228dcd7af653d Mon Sep 17 00:00:00 2001
From: Laura Blum <lblum@neb.com>
Date: Thu, 9 Apr 2026 15:46:59 -0400
Subject: [PATCH 1/7] add fgbio_clip_bam module to clip overlapping r1/r2

---
 modules/local/fgbio_clip_bam.nf    | 29 +++++++++++++++++++++++++++++
 nextflow.config                    |  1 +
 workflows/emseq_variant_calling.nf | 10 +++++++++-
 3 files changed, 39 insertions(+), 1 deletion(-)
 create mode 100644 modules/local/fgbio_clip_bam.nf

diff --git a/modules/local/fgbio_clip_bam.nf b/modules/local/fgbio_clip_bam.nf
new file mode 100644
index 0000000..998a24c
--- /dev/null
+++ b/modules/local/fgbio_clip_bam.nf
@@ -0,0 +1,29 @@
+process fgbioClipBam {
+    cpus 6
+    memory '100 GB'
+    conda "bioconda::fgbio=3.1.2 bioconda::samtools=1.19"
+    publishDir "${params.outdir}/clipped_bams/", mode: 'symlink'
+    tag { sample_id }
+
+    input:
+    tuple val(sample_id), path(bam), path(bai)
+
+    output:
+    tuple val(sample_id), path("${sample_id}.clipped.bam"), path("${sample_id}.clipped.bam.bai")
+
+    script:
+    """
+    export _JAVA_OPTIONS="-Xmx90g"
+    samtools sort -n -u -@ ${task.cpus} ${bam} | \\
+        fgbio ClipBam \\
+        --input /dev/stdin \\
+        --output /dev/stdout \\
+        --ref ${params.fasta} \\
+        --clip-overlapping-reads true \\
+        --clipping-mode Hard \\
+        --sort-order Coordinate | \\
+    samtools view -b -@ ${task.cpus} -o ${sample_id}.clipped.bam
+
+    samtools index ${sample_id}.clipped.bam
+    """
+}
diff --git a/nextflow.config b/nextflow.config
index 3b8e092..ff437ca 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -13,6 +13,7 @@ params {
     publish_dir_mode           = 'copy'
     outdir                     = 'results'
     tracedir                   = "${params.outdir}/pipeline_info"
+    run_clip_bam               = false
     run_strelka                = true
     run_freebayes              = true
     freebayes_qual_filter      = 15 // default quality filter for freebayes output.
diff --git a/workflows/emseq_variant_calling.nf b/workflows/emseq_variant_calling.nf
index 163881e..ac47033 100644
--- a/workflows/emseq_variant_calling.nf
+++ b/workflows/emseq_variant_calling.nf
@@ -33,6 +33,7 @@ happy_bed       = Channel.value(params.happy_bed ?: '') // optional regions to a
 // Local Modules:
 include {  downloadRevelio  } from '../modules/local/download_revelio.nf'
 include {  calcMD  } from '../modules/local/calc_md.nf'
+include {  fgbioClipBam  } from '../modules/local/fgbio_clip_bam.nf'
 include {  revelio  } from '../modules/local/revelio.nf'
 include {  strelka  } from '../modules/local/strelka.nf'
 include {  freebayes  } from '../modules/local/freebayes.nf'
@@ -56,11 +57,18 @@ workflow emseq_variant_calling {
         index_format
         )
 
+    //
+    // Module: Optionally clip overlapping reads with fgbio ClipBam before Revelio
+    //
+    if (params.run_clip_bam) {
+        fgbioClipBam(calcMD.out.calcmd_bam)
+    }
+
     //
     // Module: Run Revelio to mask possibly converted bases by setting BQ to 0
     //
     revelio (
-        calcMD.out.calcmd_bam,
+        params.run_clip_bam ? fgbioClipBam.out : calcMD.out.calcmd_bam,
         downloadRevelio.out,
         tmpdir
         )

From 6d6a6cbe47c98dc9ce3b97e232fa91169f39039e Mon Sep 17 00:00:00 2001
From: Laura Blum <lblum@neb.com>
Date: Wed, 15 Apr 2026 09:19:10 -0400
Subject: [PATCH 2/7] add deepvariant module with ability to customize
 checkpoint

---
 .github/workflows/ci.yml            |  5 ++++
 conf/test.config                    |  3 ++
 modules/local/deepvariant_custom.nf | 46 +++++++++++++++++++++++++++++
 nextflow.config                     |  9 ++++--
 tests/main.nf.test                  | 12 ++++++--
 tests/main.nf.test.snap             | 36 +++++++++++++++-------
 workflows/emseq_variant_calling.nf  | 29 ++++++++++++++++--
 7 files changed, 124 insertions(+), 16 deletions(-)
 create mode 100644 modules/local/deepvariant_custom.nf

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index a5767b4..7f4698d 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -39,6 +39,11 @@ jobs:
         with:
           version: 0.9.0
 
+      - name: Install Singularity
+        uses: eWaterCycle/setup-singularity@v7
+        with:
+          singularity-version: 3.8.3
+
       - name: Run pipeline with test data
         run: |
           nf-test test
diff --git a/conf/test.config b/conf/test.config
index 2c2d694..2fc7e90 100644
--- a/conf/test.config
+++ b/conf/test.config
@@ -29,4 +29,7 @@ params {
     email = "test@email.com"
     publish_dir_mode = "symlink"
     tracedir   = "test_output/pipeline_info"
+
+    // Use standard WGS model in CI (no local checkpoint available)
+    deepvariant_checkpoint = null
     }
diff --git a/modules/local/deepvariant_custom.nf b/modules/local/deepvariant_custom.nf
new file mode 100644
index 0000000..2a30519
--- /dev/null
+++ b/modules/local/deepvariant_custom.nf
@@ -0,0 +1,46 @@
+process deepvariantCustom {
+    label 'process_high'
+    tag "$library"
+    container "docker.io/google/deepvariant:1.10.0"
+    publishDir "${params.outdir}/deepvariantCustom", mode: params.publish_dir_mode
+    //
+    // Runs run_deepvariant, optionally with a fine-tuned checkpoint.
+    //
+    // When params.deepvariant_checkpoint is set, uses that checkpoint with
+    // --customized_model and --disable_small_model (required in DV 1.10.0).
+    // When null, runs with the standard built-in WGS model.
+    //
+    // The checkpoint directory must contain:
+    //   - checkpoint files (.index, .data-*)
+    //   - model.example_info.json  (required by DV 1.10.0 inference)
+    //
+    // export TMPDIR=/tmp: overrides the SGE scratch directory (TMPDIR=/scratch/JOBID)
+    //   which is not visible inside the container; parallel inside run_deepvariant
+    //   inherits this value and uses /tmp instead.
+
+    input:
+    tuple val(library), path(bam), path(bai)
+    path(fasta)
+    path(fai)
+
+    output:
+    tuple val(library), path("${library}.deepvariant.vcf.gz"),
+          path("${library}.deepvariant.vcf.gz.tbi"), emit: vcf
+
+    script:
+    def custom_model_opt = params.deepvariant_checkpoint
+        ? "--customized_model ${params.deepvariant_checkpoint} --disable_small_model"
+        : ''
+    """
+    export TMPDIR=/tmp
+
+    run_deepvariant \\
+        --model_type WGS \\
+        ${custom_model_opt} \\
+        --ref ${fasta} \\
+        --reads ${bam} \\
+        --output_vcf ${library}.deepvariant.vcf.gz \\
+        --num_shards ${task.cpus} \\
+
+    """
+}
diff --git a/nextflow.config b/nextflow.config
index ff437ca..32e3c8c 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -16,6 +16,8 @@ params {
     run_clip_bam               = false
     run_strelka                = true
     run_freebayes              = true
+    run_deepvariant            = true
+    deepvariant_checkpoint     = null
     freebayes_qual_filter      = 15 // default quality filter for freebayes output.
     run_happy                  = true
     target_bed                 = null
@@ -36,7 +38,8 @@ profiles {
         conda.cacheDir         = "${projectDir}/.conda/envs"
         conda.enabled          = true
         docker.enabled         = false
-        singularity.enabled    = false
+        singularity.enabled    = true
+        singularity.autoMounts = true
         podman.enabled         = false
         shifter.enabled        = false
         charliecloud.enabled   = false
@@ -54,10 +57,12 @@ profiles {
         apptainer.enabled      = false
     }
    
-    test { 
+    test {
         includeConfig 'conf/test.config'
         conda.enabled = true
         conda.cacheDir = "${projectDir}/.conda/envs"
+        singularity.enabled    = true
+        singularity.autoMounts = true
         }
 }
 
diff --git a/tests/main.nf.test b/tests/main.nf.test
index 2f3c844..8222681 100644
--- a/tests/main.nf.test
+++ b/tests/main.nf.test
@@ -27,6 +27,8 @@ nextflow_pipeline {
             def happy_vcf_freebayes = path("${launchDir}/test_output/happyConcordanceFreebayes/200ng-V2-1.md_vs_truth.vcf.happy.vcf.gz").vcf.summary
             def happy_summary_freebayes = path("${launchDir}/test_output/happyConcordanceFreebayes/200ng-V2-1.md_vs_truth.vcf.happy.summary.csv").text.tokenize('\n')
             def snp_breakdown = path("${launchDir}/test_output/parseHappyVcf/200ng-V2-1.md.happy.snp_mutations.txt").text.tokenize('\n')
+            def deepvariant_vcf = path("${launchDir}/test_output/deepvariantCustom/200ng-V2-1.md.deepvariant.vcf.gz").vcf.summary
+            def happy_vcf_deepvariant = path("${launchDir}/test_output/happyConcordanceDeepvariant/200ng-V2-1.md_vs_truth.vcf.happy.vcf.gz").vcf.summary
 
             assertAll(
                 { assert workflow.success },
@@ -40,7 +42,9 @@ nextflow_pipeline {
                     ["happy_summary_strelka", happy_summary_strelka],
                     ["happy_vcf_freebayes", happy_vcf_freebayes],
                     ["happy_summary_freebayes", happy_summary_freebayes],
-                    ["snp_breakdown", snp_breakdown]
+                    ["snp_breakdown", snp_breakdown],
+                    ["deepvariant_vcf", deepvariant_vcf],
+                    ["happy_vcf_deepvariant", happy_vcf_deepvariant]
                                         ).match()
                 }
             )
@@ -72,6 +76,8 @@ nextflow_pipeline {
             def happy_vcf_freebayes = path("${launchDir}/test_output/happyConcordanceFreebayes/200ng-V2-1.md_vs_truth.vcf.happy.vcf.gz").vcf.summary
             def happy_summary_freebayes = path("${launchDir}/test_output/happyConcordanceFreebayes/200ng-V2-1.md_vs_truth.vcf.happy.summary.csv").text.tokenize('\n')
             def snp_breakdown = path("${launchDir}/test_output/parseHappyVcf/200ng-V2-1.md.happy.snp_mutations.txt").text.tokenize('\n')
+            def deepvariant_vcf = path("${launchDir}/test_output/deepvariantCustom/200ng-V2-1.md.deepvariant.vcf.gz").vcf.summary
+            def happy_vcf_deepvariant = path("${launchDir}/test_output/happyConcordanceDeepvariant/200ng-V2-1.md_vs_truth.vcf.happy.vcf.gz").vcf.summary
 
             assertAll(
                 { assert workflow.success },
@@ -85,7 +91,9 @@ nextflow_pipeline {
                     ["happy_summary_strelka", happy_summary_strelka],
                     ["happy_vcf_freebayes", happy_vcf_freebayes],
                     ["happy_summary_freebayes", happy_summary_freebayes],
-                    ["snp_breakdown", snp_breakdown]
+                    ["snp_breakdown", snp_breakdown],
+                    ["deepvariant_vcf", deepvariant_vcf],
+                    ["happy_vcf_deepvariant", happy_vcf_deepvariant]
                                         ).match()
                 }
             )
diff --git a/tests/main.nf.test.snap b/tests/main.nf.test.snap
index e294762..301f7d3 100644
--- a/tests/main.nf.test.snap
+++ b/tests/main.nf.test.snap
@@ -3,8 +3,8 @@
         "content": [
             {
                 "tasksFailed": 0,
-                "tasksCount": 15,
-                "tasksSucceeded": 15
+                "tasksCount": 19,
+                "tasksSucceeded": 19
             },
             [
                 "calcmd_bam",
@@ -56,20 +56,28 @@
                     "Count Ref Alt Type",
                     "1 G C Correct"
                 ]
+            ],
+            [
+                "deepvariant_vcf",
+                "VcfFile [chromosomes=[chr21], sampleCount=1, variantCount=60, phased=false, phasedAutodetect=false]"
+            ],
+            [
+                "happy_vcf_deepvariant",
+                "VcfFile [chromosomes=[chr21], sampleCount=2, variantCount=13, phased=false, phasedAutodetect=false]"
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.04.1"
+            "nf-test": "0.9.0",
+            "nextflow": "24.04.2"
         },
-        "timestamp": "2025-05-20T13:36:17.023260625"
+        "timestamp": "2026-04-14T15:43:10.033366736"
     },
     "Pipeline test without optional inputs": {
         "content": [
             {
                 "tasksFailed": 0,
-                "tasksCount": 15,
-                "tasksSucceeded": 15
+                "tasksCount": 19,
+                "tasksSucceeded": 19
             },
             [
                 "calcmd_bam",
@@ -121,12 +129,20 @@
                     "Count Ref Alt Type",
                     "1 G C Correct"
                 ]
+            ],
+            [
+                "deepvariant_vcf",
+                "VcfFile [chromosomes=[chr21], sampleCount=1, variantCount=60, phased=false, phasedAutodetect=false]"
+            ],
+            [
+                "happy_vcf_deepvariant",
+                "VcfFile [chromosomes=[chr21], sampleCount=2, variantCount=13, phased=false, phasedAutodetect=false]"
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.04.1"
+            "nf-test": "0.9.0",
+            "nextflow": "24.04.2"
         },
-        "timestamp": "2025-05-20T13:39:05.639276115"
+        "timestamp": "2026-04-14T15:47:22.184920401"
     }
 }
\ No newline at end of file
diff --git a/workflows/emseq_variant_calling.nf b/workflows/emseq_variant_calling.nf
index ac47033..e5f57df 100644
--- a/workflows/emseq_variant_calling.nf
+++ b/workflows/emseq_variant_calling.nf
@@ -38,8 +38,10 @@ include {  revelio  } from '../modules/local/revelio.nf'
 include {  strelka  } from '../modules/local/strelka.nf'
 include {  freebayes  } from '../modules/local/freebayes.nf'
 include {  happyConcordance as happyConcordanceStrelka; 
-           happyConcordance as happyConcordanceFreebayes  } from '../modules/local/happy_concordance.nf'
+           happyConcordance as happyConcordanceFreebayes;
+           happyConcordance as happyConcordanceDeepvariant;  } from '../modules/local/happy_concordance.nf'
 include {  parseHappyVcf  } from '../modules/local/parse_happy_vcf.nf'
+include {  deepvariantCustom  } from '../modules/local/deepvariant_custom.nf'
 
 workflow emseq_variant_calling {
     
@@ -119,7 +121,7 @@ workflow emseq_variant_calling {
             fai,
             target_bed // call regions (optional)
             )
-        
+
     //
     // Module: Run hap.py to compare variants to a "truth" vcf
     //
@@ -135,6 +137,29 @@ workflow emseq_variant_calling {
             )
         }
     }
+
+    //
+    // Module: Run DeepVariant with custom model
+    //
+    if (params.run_deepvariant) {
+        deepvariantCustom(
+            revelio.out.masked,
+            fasta,
+            fai
+        )
+
+        if (params.run_happy) {
+
+            happyConcordanceDeepvariant(
+                deepvariantCustom.out.vcf,
+                happy_bed, // comparison regions (optional)
+                fasta,
+                fai,
+                happy_truth_vcf,
+                happy_truth_tbi
+            )
+        }
+    }
 }
 
 workflow {

From 8f4b4091fc3eb4e20d0cec17ec8b0f98d887a2d7 Mon Sep 17 00:00:00 2001
From: Laura Blum <lblum@neb.com>
Date: Wed, 15 Apr 2026 09:46:19 -0400
Subject: [PATCH 3/7] so that singularity can see the checkpoint folder

---
 modules/local/deepvariant_custom.nf | 10 ++++------
 workflows/emseq_variant_calling.nf  | 12 +++++++++++-
 2 files changed, 15 insertions(+), 7 deletions(-)

diff --git a/modules/local/deepvariant_custom.nf b/modules/local/deepvariant_custom.nf
index 2a30519..97dd610 100644
--- a/modules/local/deepvariant_custom.nf
+++ b/modules/local/deepvariant_custom.nf
@@ -13,23 +13,21 @@ process deepvariantCustom {
     // The checkpoint directory must contain:
     //   - checkpoint files (.index, .data-*)
     //   - model.example_info.json  (required by DV 1.10.0 inference)
-    //
-    // export TMPDIR=/tmp: overrides the SGE scratch directory (TMPDIR=/scratch/JOBID)
-    //   which is not visible inside the container; parallel inside run_deepvariant
-    //   inherits this value and uses /tmp instead.
 
     input:
     tuple val(library), path(bam), path(bai)
     path(fasta)
     path(fai)
+    path(checkpoint_dir)   // staged dir with .index/.data-*/model.example_info.json; pass [] when unused
+    val(checkpoint_name)   // checkpoint prefix e.g. "checkpoint-16896-0.97789-1"; pass '' when unused
 
     output:
     tuple val(library), path("${library}.deepvariant.vcf.gz"),
           path("${library}.deepvariant.vcf.gz.tbi"), emit: vcf
 
     script:
-    def custom_model_opt = params.deepvariant_checkpoint
-        ? "--customized_model ${params.deepvariant_checkpoint} --disable_small_model"
+    def custom_model_opt = checkpoint_name
+        ? "--customized_model ${checkpoint_dir}/${checkpoint_name} --disable_small_model"
         : ''
     """
     export TMPDIR=/tmp
diff --git a/workflows/emseq_variant_calling.nf b/workflows/emseq_variant_calling.nf
index e5f57df..91490b6 100644
--- a/workflows/emseq_variant_calling.nf
+++ b/workflows/emseq_variant_calling.nf
@@ -25,6 +25,14 @@ tmpdir = Channel.value(params.tmpdir)
 // Optional input for regions to call, e.g. for target enrichment
 target_bed = Channel.value(params.target_bed ?: '')
 
+// DeepVariant checkpoint — split single param into staged dir + name; [] and '' when unused
+checkpoint_dir  = (params.run_deepvariant && params.deepvariant_checkpoint)
+    ? Channel.value(file(params.deepvariant_checkpoint).parent)
+    : Channel.value([])
+checkpoint_name = (params.run_deepvariant && params.deepvariant_checkpoint)
+    ? Channel.value(file(params.deepvariant_checkpoint).name)
+    : Channel.value('')
+
 // Hap.py concordance inputs, only if run_happy is true
 happy_truth_vcf = params.run_happy ? Channel.fromPath(params.happy_truth_vcf, checkIfExists: true) : Channel.empty()
 happy_truth_tbi = params.run_happy ? Channel.fromPath(params.happy_truth_vcf + '.tbi', checkIfExists: true) : Channel.empty()
@@ -145,7 +153,9 @@ workflow emseq_variant_calling {
         deepvariantCustom(
             revelio.out.masked,
             fasta,
-            fai
+            fai,
+            checkpoint_dir,
+            checkpoint_name
         )
 
         if (params.run_happy) {

From b78534f884765e0114e01feb779d73d28ecd914f Mon Sep 17 00:00:00 2001
From: Laura Blum <lblum@neb.com>
Date: Fri, 17 Jul 2026 14:13:25 -0400
Subject: [PATCH 4/7] add --exome flag for targeted calling

---
 modules/local/strelka.nf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/local/strelka.nf b/modules/local/strelka.nf
index c88fe1b..847ebb6 100644
--- a/modules/local/strelka.nf
+++ b/modules/local/strelka.nf
@@ -13,7 +13,7 @@ process strelka {
     tuple val(library), path("*.strelka_vcf.gz"), path("*.strelka_vcf.gz.tbi"), emit: vcf
 
     script:
-    def target_regions_opt = (target_bed && file(target_bed).exists()) ? '--callRegions target.bed.gz' : ''
+    def target_regions_opt = (target_bed && file(target_bed).exists()) ? '--callRegions target.bed.gz --exome' : ''
     """
     if [ -n "${target_bed}" ] && [ -f "${target_bed}" ]; then
         ln -s "${target_bed}" target.bed

From 5f18748411b48c41ac8ad18bbf96daaa13e94eb5 Mon Sep 17 00:00:00 2001
From: Laura Blum <lblum@neb.com>
Date: Fri, 17 Jul 2026 17:28:19 -0400
Subject: [PATCH 5/7] add custom models for deepvariant from zenodo

---
 conf/test.config                        |  3 +-
 modules/local/deepvariant_checkpoint.nf | 66 +++++++++++++++++++++++++
 modules/local/deepvariant_custom.nf     | 16 ++----
 nextflow.config                         | 32 ++++++------
 tests/main.nf.test.snap                 | 10 ++--
 workflows/emseq_variant_calling.nf      | 13 ++---
 6 files changed, 97 insertions(+), 43 deletions(-)
 create mode 100644 modules/local/deepvariant_checkpoint.nf

diff --git a/conf/test.config b/conf/test.config
index 2fc7e90..c22bb46 100644
--- a/conf/test.config
+++ b/conf/test.config
@@ -30,6 +30,5 @@ params {
     publish_dir_mode = "symlink"
     tracedir   = "test_output/pipeline_info"
 
-    // Use standard WGS model in CI (no local checkpoint available)
-    deepvariant_checkpoint = null
+    deepvariant_model_type     = 'WGS'
     }
diff --git a/modules/local/deepvariant_checkpoint.nf b/modules/local/deepvariant_checkpoint.nf
new file mode 100644
index 0000000..472ba77
--- /dev/null
+++ b/modules/local/deepvariant_checkpoint.nf
@@ -0,0 +1,66 @@
+process fetchDeepvariantCheckpoint {
+    label 'process_single'
+    tag "${model_type}"
+    conda 'conda-forge::wget=1.20.3'
+    storeDir "${projectDir}/ref_data/deepvariant_checkpoints"
+    //
+    // Pulls the EM-seq DeepVariant fine-tuned model for the requested --model_type (only WGS and WES) from
+    // Zenodo (doi.org/10.5281/zenodo.21416823) and stages it as a checkpoint directory
+    //   WGS -> wgs.checkpoint.260717.{data-*,index}
+    //   WES -> target_capture.checkpoint.260717.{data-*,index}
+    //
+    // The output dir contains the checkpoint (.index/.data-*) plus model.example_info.json.
+    // DV 1.10.0 inference resolves it by directory: get_model_example_info_json() looks in
+    // os.path.dirname(--customized_model) for 'model.example_info.json'
+
+    input:
+    val(model_type)
+
+    output:
+    tuple val(model_type), path("${label}"), val(prefix), emit: checkpoint
+
+    script:
+    def spec = [ WGS: ['wgs',            'wgs.checkpoint.260717'],
+                 WES: ['target_capture', 'target_capture.checkpoint.260717'] ][model_type]
+    if (!spec)
+        error "No EM-seq Zenodo checkpoint for model_type '${model_type}' (only WGS and WES)."
+    label  = spec[0]
+    prefix = spec[1]
+    def base = 'https://zenodo.org/records/21416823/files'
+    """
+    mkdir -p ${label}
+
+    wget -q -O ${label}/${prefix}.data-00000-of-00001 "${base}/${prefix}.data-00000-of-00001?download=1"
+    wget -q -O ${label}/${prefix}.index               "${base}/${prefix}.index?download=1"
+    wget -q -O ${label}/model.example_info.json        "${base}/model.example_info.json?download=1"
+    """
+}
+
+workflow resolveDeepvariantCheckpoint {
+    main:
+    def mt    = params.deepvariant_model_type
+    def valid = ['WGS', 'WES', 'PACBIO', 'ONT_R104', 'HYBRID_PACBIO_ILLUMINA']
+    if (!mt)
+        error "Please set --deepvariant_model_type (one of: ${valid.join(', ')})."
+    if (!(mt in valid))
+        error "deepvariant_model_type must be one of ${valid.join(', ')}, got '${mt}'."
+    if (params.deepvariant_custom_model && !(mt in ['WGS', 'WES']))
+        error "deepvariant_custom_model is only available for WGS and WES — " +
+              "model_type '${mt}' has no custom model."
+
+    if (params.run_deepvariant == false) {
+        ch_checkpoint = Channel.value([[], '', mt])
+    } else if (params.deepvariant_checkpoint) {
+        ch_checkpoint = Channel.value([ file(params.deepvariant_checkpoint).parent,
+                                        file(params.deepvariant_checkpoint).name,
+                                        mt ])
+    } else if (params.deepvariant_custom_model) {
+        ch_checkpoint = fetchDeepvariantCheckpoint(Channel.value(mt))
+            .checkpoint.map { m, dir, prefix -> [dir, prefix, mt] }
+    } else {
+        ch_checkpoint = Channel.value([[], '', mt])
+    }
+
+    emit:
+    checkpoint = ch_checkpoint
+}
diff --git a/modules/local/deepvariant_custom.nf b/modules/local/deepvariant_custom.nf
index 97dd610..7816c34 100644
--- a/modules/local/deepvariant_custom.nf
+++ b/modules/local/deepvariant_custom.nf
@@ -3,23 +3,15 @@ process deepvariantCustom {
     tag "$library"
     container "docker.io/google/deepvariant:1.10.0"
     publishDir "${params.outdir}/deepvariantCustom", mode: params.publish_dir_mode
-    //
-    // Runs run_deepvariant, optionally with a fine-tuned checkpoint.
-    //
-    // When params.deepvariant_checkpoint is set, uses that checkpoint with
-    // --customized_model and --disable_small_model (required in DV 1.10.0).
-    // When null, runs with the standard built-in WGS model.
-    //
-    // The checkpoint directory must contain:
+    // For running a custom model the checkpoint directory must contain:
     //   - checkpoint files (.index, .data-*)
-    //   - model.example_info.json  (required by DV 1.10.0 inference)
+    //   - model.example_info.json  (required by DV 1.10.0 )
 
     input:
     tuple val(library), path(bam), path(bai)
     path(fasta)
     path(fai)
-    path(checkpoint_dir)   // staged dir with .index/.data-*/model.example_info.json; pass [] when unused
-    val(checkpoint_name)   // checkpoint prefix e.g. "checkpoint-16896-0.97789-1"; pass '' when unused
+    tuple path(checkpoint_dir), val(checkpoint_name), val(model_type)
 
     output:
     tuple val(library), path("${library}.deepvariant.vcf.gz"),
@@ -33,7 +25,7 @@ process deepvariantCustom {
     export TMPDIR=/tmp
 
     run_deepvariant \\
-        --model_type WGS \\
+        --model_type ${model_type} \\
         ${custom_model_opt} \\
         --ref ${fasta} \\
         --reads ${bam} \\
diff --git a/nextflow.config b/nextflow.config
index 32e3c8c..ad2ac0f 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -8,21 +8,23 @@
 
 // Global default params, used in configs
 params {
-    tmpdir                     = '/tmp'
-    index_format               = 'csi'
-    publish_dir_mode           = 'copy'
-    outdir                     = 'results'
-    tracedir                   = "${params.outdir}/pipeline_info"
-    run_clip_bam               = false
-    run_strelka                = true
-    run_freebayes              = true
-    run_deepvariant            = true
-    deepvariant_checkpoint     = null
-    freebayes_qual_filter      = 15 // default quality filter for freebayes output.
-    run_happy                  = true
-    target_bed                 = null
-    happy_truth_vcf            = null
-    happy_bed                  = null
+    tmpdir                       = '/tmp'
+    index_format                 = 'csi'
+    publish_dir_mode             = 'copy'
+    outdir                       = 'results'
+    tracedir                     = "${params.outdir}/pipeline_info"
+    run_clip_bam                 = false
+    run_strelka                  = true
+    run_freebayes                = true
+    run_deepvariant              = true
+    deepvariant_model_type       = null  // WGS | WES | PACBIO | ONT_R104 | HYBRID_PACBIO_ILLUMINA
+    deepvariant_custom_model     = true  // Download an EM-seq fine-tuned model, only available for WGS and WES (target capture).
+    deepvariant_checkpoint       = null  // Optional local custom model prefix
+    freebayes_qual_filter        = 15
+    run_happy                    = true
+    target_bed                   = null
+    happy_truth_vcf              = null
+    happy_bed                    = null
 }
 
 includeConfig 'conf/base.config'
diff --git a/tests/main.nf.test.snap b/tests/main.nf.test.snap
index 301f7d3..b2177d0 100644
--- a/tests/main.nf.test.snap
+++ b/tests/main.nf.test.snap
@@ -54,7 +54,7 @@
                 "snp_breakdown",
                 [
                     "Count Ref Alt Type",
-                    "1 G C Correct"
+                    "1 G C NotPassed"
                 ]
             ],
             [
@@ -63,14 +63,14 @@
             ],
             [
                 "happy_vcf_deepvariant",
-                "VcfFile [chromosomes=[chr21], sampleCount=2, variantCount=13, phased=false, phasedAutodetect=false]"
+                "VcfFile [chromosomes=[chr21], sampleCount=2, variantCount=4, phased=false, phasedAutodetect=false]"
             ]
         ],
         "meta": {
             "nf-test": "0.9.0",
             "nextflow": "24.04.2"
         },
-        "timestamp": "2026-04-14T15:43:10.033366736"
+        "timestamp": "2026-07-17T17:08:12.351084775"
     },
     "Pipeline test without optional inputs": {
         "content": [
@@ -136,13 +136,13 @@
             ],
             [
                 "happy_vcf_deepvariant",
-                "VcfFile [chromosomes=[chr21], sampleCount=2, variantCount=13, phased=false, phasedAutodetect=false]"
+                "VcfFile [chromosomes=[chr21], sampleCount=2, variantCount=4, phased=false, phasedAutodetect=false]"
             ]
         ],
         "meta": {
             "nf-test": "0.9.0",
             "nextflow": "24.04.2"
         },
-        "timestamp": "2026-04-14T15:47:22.184920401"
+        "timestamp": "2026-07-17T17:11:27.137540031"
     }
 }
\ No newline at end of file
diff --git a/workflows/emseq_variant_calling.nf b/workflows/emseq_variant_calling.nf
index 91490b6..d430aea 100644
--- a/workflows/emseq_variant_calling.nf
+++ b/workflows/emseq_variant_calling.nf
@@ -25,13 +25,7 @@ tmpdir = Channel.value(params.tmpdir)
 // Optional input for regions to call, e.g. for target enrichment
 target_bed = Channel.value(params.target_bed ?: '')
 
-// DeepVariant checkpoint — split single param into staged dir + name; [] and '' when unused
-checkpoint_dir  = (params.run_deepvariant && params.deepvariant_checkpoint)
-    ? Channel.value(file(params.deepvariant_checkpoint).parent)
-    : Channel.value([])
-checkpoint_name = (params.run_deepvariant && params.deepvariant_checkpoint)
-    ? Channel.value(file(params.deepvariant_checkpoint).name)
-    : Channel.value('')
+// DeepVariant checkpoint — resolved to (dir, name): local override, else Zenodo fetch by mode.
 
 // Hap.py concordance inputs, only if run_happy is true
 happy_truth_vcf = params.run_happy ? Channel.fromPath(params.happy_truth_vcf, checkIfExists: true) : Channel.empty()
@@ -50,6 +44,7 @@ include {  happyConcordance as happyConcordanceStrelka;
            happyConcordance as happyConcordanceDeepvariant;  } from '../modules/local/happy_concordance.nf'
 include {  parseHappyVcf  } from '../modules/local/parse_happy_vcf.nf'
 include {  deepvariantCustom  } from '../modules/local/deepvariant_custom.nf'
+include {  resolveDeepvariantCheckpoint  } from '../modules/local/deepvariant_checkpoint.nf'
 
 workflow emseq_variant_calling {
     
@@ -150,12 +145,12 @@ workflow emseq_variant_calling {
     // Module: Run DeepVariant with custom model
     //
     if (params.run_deepvariant) {
+        ck = resolveDeepvariantCheckpoint()
         deepvariantCustom(
             revelio.out.masked,
             fasta,
             fai,
-            checkpoint_dir,
-            checkpoint_name
+            ck.checkpoint
         )
 
         if (params.run_happy) {

From a94690c7082ee700b012c7d6190c9f3a555284c0 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Fri, 17 Jul 2026 21:41:45 +0000
Subject: [PATCH 6/7] Fix CI: use Docker in test profile, update NXF_VER, fix
 trailing backslash in deepvariant script

---
 .github/workflows/ci.yml            | 7 +------
 modules/local/deepvariant_custom.nf | 5 ++---
 nextflow.config                     | 4 ++--
 3 files changed, 5 insertions(+), 11 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 7f4698d..9c1f46a 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -13,7 +13,7 @@ jobs:
     strategy:
       matrix:
         NXF_VER:
-          - "23.10.0"
+          - "25.04.2"
     steps:
       - name: Check out pipeline code
         uses: actions/checkout@v4
@@ -39,11 +39,6 @@ jobs:
         with:
           version: 0.9.0
 
-      - name: Install Singularity
-        uses: eWaterCycle/setup-singularity@v7
-        with:
-          singularity-version: 3.8.3
-
       - name: Run pipeline with test data
         run: |
           nf-test test
diff --git a/modules/local/deepvariant_custom.nf b/modules/local/deepvariant_custom.nf
index 7816c34..d994e5e 100644
--- a/modules/local/deepvariant_custom.nf
+++ b/modules/local/deepvariant_custom.nf
@@ -30,7 +30,6 @@ process deepvariantCustom {
         --ref ${fasta} \\
         --reads ${bam} \\
         --output_vcf ${library}.deepvariant.vcf.gz \\
-        --num_shards ${task.cpus} \\
-
-    """
+        --num_shards ${task.cpus}
+"""
 }
diff --git a/nextflow.config b/nextflow.config
index ad2ac0f..d4c9a06 100644
--- a/nextflow.config
+++ b/nextflow.config
@@ -63,8 +63,8 @@ profiles {
         includeConfig 'conf/test.config'
         conda.enabled = true
         conda.cacheDir = "${projectDir}/.conda/envs"
-        singularity.enabled    = true
-        singularity.autoMounts = true
+        docker.enabled = true
+        singularity.enabled = false
         }
 }
 

From b6465e5865c5f6fd8a43ebae9a05e005582d3683 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Fri, 17 Jul 2026 22:04:04 +0000
Subject: [PATCH 7/7] Fix CI: update snapshot tasksCount from 19 to 20 for
 small dataset test

---
 tests/main.nf.test.snap | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/main.nf.test.snap b/tests/main.nf.test.snap
index b2177d0..78f1080 100644
--- a/tests/main.nf.test.snap
+++ b/tests/main.nf.test.snap
@@ -3,8 +3,8 @@
         "content": [
             {
                 "tasksFailed": 0,
-                "tasksCount": 19,
-                "tasksSucceeded": 19
+                "tasksCount": 20,
+                "tasksSucceeded": 20
             },
             [
                 "calcmd_bam",