From 672add6f643d5f3306437671f23509eeec8a5a3c Mon Sep 17 00:00:00 2001
From: Danielle Callan <dmleake92@gmail.com>
Date: Thu, 9 Jan 2025 14:25:34 -0500
Subject: [PATCH] feat: cleaning up and adding more workflows (#206)

---
 files/source/assemblies.yml       | 22 ++++++++++++++++++
 files/source/galaxy_workflows.yml | 37 ++++++++++++++++++++++++++++---
 2 files changed, 56 insertions(+), 3 deletions(-)

diff --git a/files/source/assemblies.yml b/files/source/assemblies.yml
index 05026f4..7576f57 100644
--- a/files/source/assemblies.yml
+++ b/files/source/assemblies.yml
@@ -1,70 +1,92 @@
 assemblies:
+  # plasmodium falciparum
   - accession: GCF_000002765.6
     ploidy: haploid
     source: plasmodb.org
+  # plasmodium vivax
   - accession: GCF_000002415.2
     ploidy: haploid
     ncbi_dataset: true
   - accession: GCA_900093555.2
     ploidy: haploid
     source: plasmodb.org
+  # plasmodium yoelii
   - accession: GCF_900002385.2
     ploidy: haploid
     source: plasmodb.org
+  # plasmodium vinckei
   - accession: GCF_900681995.1
     ploidy: haploid
     source: plasmodb.org
+  # culex p. pallens
   - accession: GCF_016801865.2
     ploidy: diploid
     source: vectorbase.org
+  # culex p. quinquefasciatus
   - accession: GCF_015732765.1
     ploidy: diploid
     source: vectorbase.org
+  # anopheles gambiae
   - accession: GCF_943734735.2
     ploidy: diploid
     ncbi_dataset: true
   - accession: GCF_000005575.2
     ploidy: diploid
     source: vectorbase.org
+  # toxoplasma gondii
   - accession: GCF_000006565.2
     ploidy: haploid
     source: toxodb.org
+  # mycobacterium tuberculosis
   - accession: GCF_000195955.2
     ploidy: haploid
+  # coccidioides posadasii
   - accession: GCF_018416015.2
     ploidy: haploid
     ncbi_dataset: true
   - accession: GCA_000170175.2
     ploidy: haploid
     source: fungidb.org
+  # coccidioides immitis
   - accession: GCF_000149335.2
     ploidy: haploid
     source: fungidb.org
+  # trypanosoma cruzi
   - accession: GCF_000209065.1
     ploidy: diploid
     source: tritrypdb.org
+  # trypanosoma brucei
   - accession: GCF_000002445.2
     ploidy: diploid
     source: tritrypdb.org
+  # leishmania major
   - accession: GCF_000002725.2
     ploidy: diploid
     source: tritrypdb.org
+  # leishmania donovani
   - accession: GCF_000227135.1
     ploidy: diploid
     source: tritrypdb.org
+  # leishmania braziliensis
+  # danielle's favorite
   - accession: GCF_000002845.2
     ploidy: diploid
     source: tritrypdb.org
+  # covid
   - accession: GCF_009858895.2
     ploidy: haploid
+  # mpox
   - accession: GCF_000857045.1
     ploidy: haploid
+  # aspergillus fumigatus
   - accession: GCF_000002655.1
     ploidy: haploid
     source: fungidb.org
+  # candida albicans
   - accession: GCF_000182965.3
     ploidy: diploid
     source: fungidb.org
+  # crypococcus neoformans
   - accession: GCF_000091045.1
     ploidy: haploid
     ncbi_dataset: true
diff --git a/files/source/galaxy_workflows.yml b/files/source/galaxy_workflows.yml
index 351f06f..a39efc8 100644
--- a/files/source/galaxy_workflows.yml
+++ b/files/source/galaxy_workflows.yml
@@ -1,15 +1,46 @@
 workflows:
-  - trs_id: "https://dockstore.org/api/ga4gh/trs/v2/tools/#workflow/github.com/iwc-workflows/rnaseq-pe/main/versions/v0.9"
+  # this list of workflows and their annotations are a wip and will be updated as i learn more
+  # about their status and the tools used, etc.
+  - trs_id: "#workflow/github.com/iwc-workflows/chipseq-pe/main"
     type: "REGULATION"
     description: "Workflows for the analysis of ChIP-seq, ATAC-Seq, and beyond."
     ploidy: "any"
 
-  - trs_id: "https://dockstore.org/api/ga4gh/trs/v2/tools/#workflow/github.com/iwc-workflows/chipseq-pe/main/versions/v0.12"
+  - trs_id: "#workflow/github.com/iwc-workflows/chipseq-sr/main"
+    type: "REGULATION"
+    description: "Workflows for the analysis of ChIP-seq, ATAC-Seq, and beyond."
+    ploidy: "any"
+
+  - trs_id: "#workflow/github.com/iwc-workflows/rnaseq-pe/main"
+    type: "TRANSCRIPTOMICS"
+    description: "Analyze bulk or single-cell RNA seq data using a variety of approaches."
+    ploidy: "any"
+
+  - trs_id: "#workflow/github.com/iwc-workflows/rnaseq-sr/main"
     type: "TRANSCRIPTOMICS"
     description: "Analyze bulk or single-cell RNA seq data using a variety of approaches."
     ploidy: "any"
 
-  - trs_id: "https://dockstore.org/api/ga4gh/trs/v2/tools/#workflow/github.com/iwc-workflows/haploid-variant-calling-wgs-pe/main/versions/v0.1"
+  - trs_id: "#workflow/github.com/iwc-workflows/haploid-variant-calling-wgs-pe/main"
     type: "VARIANT_CALLING"
     description: "Identify nucleotide polymorphisms and short indels from Illumina and Element data."
     ploidy: "haploid"
+
+  # im starting to wonder looking at this if the default for ploidy shouldnt be any
+  # maybe it could be multivalued [haploid, diploid].. might not matter depending on ui
+  - trs_id: "#workflow/github.com/iwc-workflows/generic-variant-calling-wgs-pe/main"
+    type: "VARIANT_CALLING"
+    description: "Identify nucleotide polymorphisms and short indels from Illumina and Element data."
+    ploidy: "any"
+
+  - trs_id: "#workflow/github.com/iwc-workflows/assembly-with-flye/main"
+    type: "ASSEMBLY"
+    description: "Best practices for assembly of prokaryotic and eukaryotic genomes sequenced with a variety of technologies."
+    ploidy: "any"
+
+  # idk yet what makes this specific to bacteria, or if it can be easily generalized
+  # but we may need a new annotation once i figure it out
+  - trs_id: "#workflow/github.com/iwc-workflows/bacterial-genome-assembly/main"
+    type: "ASSEMBLY"
+    description: "Best practices for assembly of prokaryotic and eukaryotic genomes sequenced with a variety of technologies."
+    ploidy: "any"