chore: Synchronize pipeline validation scripts across submodule metrics

Author: AI Agent

config/amalgkit/amalgkit_acromyrmex_echinatior.yaml

@@ -11,8 +11,8 @@
 # Generated: 2026-02-09
 
 # Paths are resolved relative to repository root
-work_dir: output/amalgkit/acromyrmex_echinatior/work
-log_dir: output/amalgkit/acromyrmex_echinatior/logs
+work_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/work
+log_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/logs
 threads: 16
 
 auto_install_amalgkit: true
@@ -32,7 +32,7 @@ genome:
   accession: GCF_000204515.1
   assembly_name: Aech_3.9
   annotation_release: 100
-  dest_dir: output/amalgkit/shared/genome/Acromyrmex_echinatior
+  dest_dir: projects/hymenoptera_amalgkit/data/shared/genome/Acromyrmex_echinatior
   include:
     - genome
     - gff3
@@ -50,18 +50,18 @@ genome:
 # Per-step parameters
 steps:
   metadata:
-    out_dir: output/amalgkit/acromyrmex_echinatior/work
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/work
     search_string: '"Acromyrmex echinatior"[Organism] AND "RNA-Seq"[Strategy] AND "Illumina"[Platform]'
     redo: no
   integrate:
-    fastq_dir: output/amalgkit/acromyrmex_echinatior/fastq/getfastq
+    fastq_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/fastq/getfastq
   select:
     max_sample: 999999
     mark_missing_rank: none
     priority: "completeness"
     redo: no
   getfastq:
-    out_dir: output/amalgkit/acromyrmex_echinatior/fastq
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/fastq
     remove_sra: yes
     threads: 16
     jobs: 16
@@ -73,22 +73,22 @@ steps:
     fastp: no
   quant:
     jobs: 1
-    out_dir: output/amalgkit/acromyrmex_echinatior/work
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/work
     threads: 16
     redo: no
     build_index: no
     keep_fastq: no
-    index_dir: output/amalgkit/shared/genome/Acromyrmex_echinatior/index
-    fasta_dir: output/amalgkit/shared/genome/Acromyrmex_echinatior
+    index_dir: projects/hymenoptera_amalgkit/data/shared/genome/Acromyrmex_echinatior/index
+    fasta_dir: projects/hymenoptera_amalgkit/data/shared/genome/Acromyrmex_echinatior
   merge:
-    out: output/amalgkit/acromyrmex_echinatior/merged/merged_abundance.tsv
-    out_dir: output/amalgkit/acromyrmex_echinatior/merged
+    out: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/merged/merged_abundance.tsv
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/merged
   cstmm:
-    out_dir: output/amalgkit/acromyrmex_echinatior/work
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/work
     orthogroup_table: ""
   curate:
-    out_dir: output/amalgkit/acromyrmex_echinatior/work
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/work
   csca:
-    out_dir: output/amalgkit/acromyrmex_echinatior/work
+    out_dir: projects/hymenoptera_amalgkit/data/acromyrmex_echinatior/work
     orthogroup_table: ""
   sanity: {}

config/amalgkit/amalgkit_anoplolepis_gracilipes.yaml

@@ -11,8 +11,8 @@
 # Generated: 2026-02-09
 
 # Paths are resolved relative to repository root
-work_dir: output/amalgkit/anoplolepis_gracilipes/work
-log_dir: output/amalgkit/anoplolepis_gracilipes/logs
+work_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/work
+log_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/logs
 threads: 16
 
 auto_install_amalgkit: true
@@ -32,7 +32,7 @@ genome:
   accession: GCF_047496725.1
   assembly_name: ASM4749672v1
   annotation_release: RS_2025_04
-  dest_dir: output/amalgkit/shared/genome/Anoplolepis_gracilipes
+  dest_dir: projects/hymenoptera_amalgkit/data/shared/genome/Anoplolepis_gracilipes
   include:
     - genome
     - gff3
@@ -50,18 +50,18 @@ genome:
 # Per-step parameters
 steps:
   metadata:
-    out_dir: output/amalgkit/anoplolepis_gracilipes/work
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/work
     search_string: '"Anoplolepis gracilipes"[Organism] AND "RNA-Seq"[Strategy] AND "Illumina"[Platform]'
     redo: no
   integrate:
-    fastq_dir: output/amalgkit/anoplolepis_gracilipes/fastq/getfastq
+    fastq_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/fastq/getfastq
   select:
     max_sample: 999999
     mark_missing_rank: none
     priority: "completeness"
     redo: no
   getfastq:
-    out_dir: output/amalgkit/anoplolepis_gracilipes/fastq
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/fastq
     threads: 16
     jobs: 16
     redo: no
@@ -73,22 +73,22 @@ steps:
     max_bp: 50000000
   quant:
     jobs: 1
-    out_dir: output/amalgkit/anoplolepis_gracilipes/work
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/work
     threads: 16
     redo: no
     build_index: no
     keep_fastq: no
-    index_dir: output/amalgkit/shared/genome/Anoplolepis_gracilipes/index
-    fasta_dir: output/amalgkit/shared/genome/Anoplolepis_gracilipes
+    index_dir: projects/hymenoptera_amalgkit/data/shared/genome/Anoplolepis_gracilipes/index
+    fasta_dir: projects/hymenoptera_amalgkit/data/shared/genome/Anoplolepis_gracilipes
   merge:
-    out: output/amalgkit/anoplolepis_gracilipes/merged/merged_abundance.tsv
-    out_dir: output/amalgkit/anoplolepis_gracilipes/merged
+    out: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/merged/merged_abundance.tsv
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/merged
   cstmm:
-    out_dir: output/amalgkit/anoplolepis_gracilipes/work
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/work
     orthogroup_table: ""
   curate:
-    out_dir: output/amalgkit/anoplolepis_gracilipes/work
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/work
   csca:
-    out_dir: output/amalgkit/anoplolepis_gracilipes/work
+    out_dir: projects/hymenoptera_amalgkit/data/anoplolepis_gracilipes/work
     orthogroup_table: ""
   sanity: {}

config/amalgkit/amalgkit_apis_mellifera.yaml

@@ -7,8 +7,8 @@
 # Notes: Model organism for social behavior, extensive RNA-seq datasets available
 
 
-work_dir: output/amalgkit/apis_mellifera/work
-log_dir: output/amalgkit/apis_mellifera/logs
+work_dir: projects/hymenoptera_amalgkit/data/apis_mellifera/work
+log_dir: projects/hymenoptera_amalgkit/data/apis_mellifera/logs
 threads: 60
 
 
@@ -35,7 +35,7 @@ genome:
   accession: GCF_003254395.2
   assembly_name: Amel_HAv3.1
   annotation_release: 104
-  dest_dir: output/amalgkit/apis_mellifera/genome
+  dest_dir: projects/hymenoptera_amalgkit/data/apis_mellifera/genome
 
 
   # Files to download (based on available NCBI FTP resources)
@@ -79,7 +79,7 @@ steps:
     search_string: '"Apis mellifera"[Organism] AND RNA-Seq[Strategy] AND Illumina[Platform]'
     redo: no
   integrate:
-    fastq_dir: output/amalgkit/apis_mellifera/fastq/getfastq
+    fastq_dir: projects/hymenoptera_amalgkit/data/apis_mellifera/fastq/getfastq
   config: {}
   select: {}
   getfastq:
@@ -96,7 +96,7 @@ steps:
     # Delete SRA/FASTQ files after quantification to save space
     keep_fastq: no
   merge:
-    out: output/amalgkit/apis_mellifera/merged/merged_abundance.tsv
+    out: projects/hymenoptera_amalgkit/data/apis_mellifera/merged/merged_abundance.tsv
   cstmm: {}
   curate: {}
   csca: {}

config/amalgkit/amalgkit_athalia_rosae.yaml

@@ -4,8 +4,8 @@
 # Assembly: GCF_000515845.1
 # Note: Added for Hymenoptera cross-species tissue comparisons (antenna, brain, ovary, tubules)
 
-work_dir: output/amalgkit/athalia_rosae/work
-log_dir: output/amalgkit/athalia_rosae/logs
+work_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/work
+log_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/logs
 threads: 16
 
 auto_install_amalgkit: true
@@ -20,7 +20,7 @@ taxon_id: 37344
 
 genome:
   accession: GCF_000515845.1
-  dest_dir: output/amalgkit/athalia_rosae/genome
+  dest_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/genome
 
   include:
     - genome           
@@ -39,32 +39,32 @@ genome:
 
 steps:
   metadata:
-    out_dir: output/amalgkit/athalia_rosae/work
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/work
     search_string: '"Athalia rosae"[Organism] AND RNA-Seq[Strategy] AND Illumina[Platform]'
     redo: no
   integrate:
-    fastq_dir: output/amalgkit/athalia_rosae/fastq/getfastq
+    fastq_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/fastq/getfastq
   config: {}
   select: {}
   getfastq:
-    out_dir: output/amalgkit/athalia_rosae/fastq
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/fastq
     remove_sra: yes
     threads: 16
     pfd: yes
     redo: no
   quant:
     jobs: 1
-    out_dir: output/amalgkit/athalia_rosae/quant
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/quant
     threads: 16
     redo: no
     keep_fastq: no
   merge:
-    out: output/amalgkit/athalia_rosae/merged/merged_abundance.tsv
-    out_dir: output/amalgkit/athalia_rosae/merged
+    out: projects/hymenoptera_amalgkit/data/athalia_rosae/merged/merged_abundance.tsv
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/merged
   cstmm:
-    out_dir: output/amalgkit/athalia_rosae/cstmm
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/cstmm
   curate:
-    out_dir: output/amalgkit/athalia_rosae/curate
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/curate
   csca:
-    out_dir: output/amalgkit/athalia_rosae/csca
+    out_dir: projects/hymenoptera_amalgkit/data/athalia_rosae/csca
   sanity: {}

config/amalgkit/amalgkit_atta_cephalotes.yaml

@@ -11,8 +11,8 @@
 # Generated: 2026-02-09
 
 # Paths are resolved relative to repository root
-work_dir: output/amalgkit/atta_cephalotes/work
-log_dir: output/amalgkit/atta_cephalotes/logs
+work_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/work
+log_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/logs
 threads: 16
 
 auto_install_amalgkit: true
@@ -32,7 +32,7 @@ genome:
   accession: GCF_000143395.1
   assembly_name: Attacep1.0
   annotation_release: 100
-  dest_dir: output/amalgkit/shared/genome/Atta_cephalotes
+  dest_dir: projects/hymenoptera_amalgkit/data/shared/genome/Atta_cephalotes
   include:
     - genome
     - gff3
@@ -50,18 +50,18 @@ genome:
 # Per-step parameters
 steps:
   metadata:
-    out_dir: output/amalgkit/atta_cephalotes/work
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/work
     search_string: '"Atta cephalotes"[Organism] AND "RNA-Seq"[Strategy] AND "Illumina"[Platform]'
     redo: no
   integrate:
-    fastq_dir: output/amalgkit/atta_cephalotes/fastq/getfastq
+    fastq_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/fastq/getfastq
   select:
     max_sample: 999999
     mark_missing_rank: none
     priority: "completeness"
     redo: no
   getfastq:
-    out_dir: output/amalgkit/atta_cephalotes/fastq
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/fastq
     threads: 16
     jobs: 16
     redo: no
@@ -73,22 +73,22 @@ steps:
     max_bp: 50000000
   quant:
     jobs: 1
-    out_dir: output/amalgkit/atta_cephalotes/work
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/work
     threads: 16
     redo: no
     build_index: no
     keep_fastq: no
-    index_dir: output/amalgkit/shared/genome/Atta_cephalotes/index
-    fasta_dir: output/amalgkit/shared/genome/Atta_cephalotes
+    index_dir: projects/hymenoptera_amalgkit/data/shared/genome/Atta_cephalotes/index
+    fasta_dir: projects/hymenoptera_amalgkit/data/shared/genome/Atta_cephalotes
   merge:
-    out: output/amalgkit/atta_cephalotes/merged/merged_abundance.tsv
-    out_dir: output/amalgkit/atta_cephalotes/merged
+    out: projects/hymenoptera_amalgkit/data/atta_cephalotes/merged/merged_abundance.tsv
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/merged
   cstmm:
-    out_dir: output/amalgkit/atta_cephalotes/work
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/work
     orthogroup_table: ""
   curate:
-    out_dir: output/amalgkit/atta_cephalotes/work
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/work
   csca:
-    out_dir: output/amalgkit/atta_cephalotes/work
+    out_dir: projects/hymenoptera_amalgkit/data/atta_cephalotes/work
     orthogroup_table: ""
   sanity: {}

config/amalgkit/amalgkit_bombus_terrestris.yaml

@@ -4,8 +4,8 @@
 # Assembly: GCF_910591885.1 (iyBomTerr1.2)
 # Note: Added for Hymenoptera cross-species tissue comparisons (bumblebee)
 
-work_dir: output/amalgkit/bombus_terrestris/work
-log_dir: output/amalgkit/bombus_terrestris/logs
+work_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/work
+log_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/logs
 threads: 16
 
 auto_install_amalgkit: true
@@ -25,7 +25,7 @@ taxon_id: 30195
 genome:
   accession: GCF_910591885.1
   assembly_name: iyBomTerr1.2
-  dest_dir: output/amalgkit/bombus_terrestris/genome
+  dest_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/genome
 
   # Files to download (based on available NCBI FTP resources)
   include:
@@ -49,32 +49,32 @@ genome:
 # Per-step parameters (merged on top of common params)
 steps:
   metadata:
-    out_dir: output/amalgkit/bombus_terrestris/work
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/work
     search_string: '"Bombus terrestris"[Organism] AND RNA-Seq[Strategy] AND Illumina[Platform]'
     redo: no
   integrate:
-    fastq_dir: output/amalgkit/bombus_terrestris/fastq/getfastq
+    fastq_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/fastq/getfastq
   config: {}
   select: {}
   getfastq:
-    out_dir: output/amalgkit/bombus_terrestris/fastq
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/fastq
     remove_sra: yes
     threads: 16
     pfd: yes
     redo: no
   quant:
     jobs: 1
-    out_dir: output/amalgkit/bombus_terrestris/quant
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/quant
     threads: 16
     redo: no
     keep_fastq: no
   merge:
-    out: output/amalgkit/bombus_terrestris/merged/merged_abundance.tsv
-    out_dir: output/amalgkit/bombus_terrestris/merged
+    out: projects/hymenoptera_amalgkit/data/bombus_terrestris/merged/merged_abundance.tsv
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/merged
   cstmm:
-    out_dir: output/amalgkit/bombus_terrestris/cstmm
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/cstmm
   curate:
-    out_dir: output/amalgkit/bombus_terrestris/curate
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/curate
   csca:
-    out_dir: output/amalgkit/bombus_terrestris/csca
+    out_dir: projects/hymenoptera_amalgkit/data/bombus_terrestris/csca
   sanity: {}