From 53579e15b6357d114eb7da11582211fbaca26df7 Mon Sep 17 00:00:00 2001
From: Christopher Tomkins-Tinch <tomkinsc@broadinstitute.org>
Date: Wed, 18 Sep 2024 23:54:41 -0400
Subject: [PATCH] initial WIP commit of a task and workflow to run delphy

initial WIP commit of a task and workflow to run delphy (https://github.com/broadinstitute/delphy). Further work pending https://github.com/broadinstitute/delphy/pull/2 and availability of a delphy docker image
---
 .dockstore.yml                       |  5 ++
 pipes/WDL/tasks/tasks_delphy.wdl     | 92 ++++++++++++++++++++++++++++
 pipes/WDL/workflows/delphy_phylo.wdl | 23 +++++++
 requirements-modules.txt             |  1 +
 4 files changed, 121 insertions(+)
 create mode 100644 pipes/WDL/tasks/tasks_delphy.wdl
 create mode 100644 pipes/WDL/workflows/delphy_phylo.wdl

diff --git a/.dockstore.yml b/.dockstore.yml
index 0c9012485..3bb666675 100644
--- a/.dockstore.yml
+++ b/.dockstore.yml
@@ -100,6 +100,11 @@ workflows:
     primaryDescriptorPath: /pipes/WDL/workflows/coverage_table.wdl
     testParameterFiles:
       - /empty.json
+  - name: delphy_phylo
+    subclass: WDL
+    primaryDescriptorPath: /pipes/WDL/workflows/delphy_phylo.wdl
+    testParameterFiles:
+      - /empty.json
   - name: demux_metag
     subclass: WDL
     primaryDescriptorPath: /pipes/WDL/workflows/demux_metag.wdl
diff --git a/pipes/WDL/tasks/tasks_delphy.wdl b/pipes/WDL/tasks/tasks_delphy.wdl
new file mode 100644
index 000000000..c17efff71
--- /dev/null
+++ b/pipes/WDL/tasks/tasks_delphy.wdl
@@ -0,0 +1,92 @@
+task delphy {
+  input {
+    File sequences_msa_fasta
+
+    Int? num_steps
+    Int? tree_every_n_steps
+    Int? snapshot_every_n_steps
+    Int? log_every_n_steps
+
+    Int? num_threads # min($(nproc),(num_sequences / 100)) suggested
+
+    Int? cpus
+    Int? machine_mem_gb
+    Int disk_dize = 300
+
+    String docker = "quay.io/broadinstitute/delphy:0.999"
+  }
+
+  String out_basename = basename(basename(sequences_msa_fasta,".fa"),".fasta")
+
+  meta {
+    description: "Execute Delphy; see: https://github.com/broadinstitute/delphy"
+  }
+  parameter_meta {
+    sequences_msa_fasta: {
+      description: "multiple alignment of input sequences in fasta format"
+    }
+    num_steps: {
+      description: "(5000000 * num_sequences) suggested"
+    }
+    tree_every_n_steps: {
+      description: "(num_steps / 200) suggested"
+    }
+    snapshot_every_n_steps: {
+      description: "(num_steps / 200) suggested"
+    }
+    log_every_n_steps: {
+      description: "(num_steps / 10000) suggested"
+    }
+    num_threads: {
+      description: "min($(nproc),(num_sequences / 100)) suggested"
+    }
+  }
+
+  command <<<
+    set -e
+    delphy --version > DELPHY_VERSION
+
+    num_sequences=$(grep -c ">" ~{sequences_msa_fasta})
+
+    steps=~{if defined(num_steps) then "~{num_steps}" else "$((($num_sequences * 5000000)))"}
+    tree_every=~{if defined(tree_every_n_steps) then "~{tree_every_n_steps}" else "$((($steps / 200)))"}
+    snapshot_every=~{if defined(snapshot_every_n_steps) then "~{snapshot_every_n_steps}" else "$((($steps / 200)))"}
+    log_every=~{if defined(log_every_n_steps) then "~{log_every_n_steps}" else "$((($steps / 10000)))"}
+
+    threads=~{if defined(num_threads) then "~{num_threads}" else "$(nproc)"}
+
+    delphy \
+       --v0-in-fasta ~{sequences_msa_fasta} \
+       --v0-threads $threads \
+       #[--v0-site-rate-heterogeneity] \
+       --v0-steps $steps \
+       --v0-tree-every $tree_every \
+       --v0-delphy-snapshot-every $snapshot_every \
+       --v0-log-every $log_every \
+       --v0-out-trees-file  "~{out_basename}.dphy.trees" \
+       --v0-out-log-file    "~{out_basename}.dphy.log" \
+       --v0-out-delphy-file "~{out_basename}.dphy" \
+       --v0-out-beast-xml   "~{out_basename}.dphy.for_beast.xml"
+  >>>
+
+  output {
+    File   delphy_trees            = "~{out_basename}.dphy.trees"
+    File   delphy_log              = "~{out_basename}.dphy.log"
+    File   delphy_for_web_ui       = "~{out_basename}.dphy"
+    File   delphy_output_for_beast = "~{out_basename}.dphy.for_beast.xml"
+    File   delphy_stdout           = stdout()
+    
+    String delphy_version          = read_string('DELPHY_VERSION')
+  }
+
+  runtime {
+    docker:           docker
+    memory:           select_first([machine_mem_gb, 15]) + " GB"
+    cpu:              select_first([cpus, 8])
+    disks:            "local-disk " + disk_size + " HDD"
+    disk:             disk_size + " GB" # TES
+    dx_instance_type: "mem1_ssd1_gpu2_x8" # dxWDL
+
+    maxRetries: 1
+  }
+}
\ No newline at end of file
diff --git a/pipes/WDL/workflows/delphy_phylo.wdl b/pipes/WDL/workflows/delphy_phylo.wdl
new file mode 100644
index 000000000..9022239ac
--- /dev/null
+++ b/pipes/WDL/workflows/delphy_phylo.wdl
@@ -0,0 +1,23 @@
+version 1.0
+
+import "../tasks/tasks_delphy.wdl" as delphy
+
+workflow delphy_phylo {
+    meta {
+        description: "Runs Delphy"
+        author:      "Broad Viral Genomics"
+        email:       "viral-ngs@broadinstitute.org"
+    }
+
+    call delphy.delphy
+
+    output {
+        File   delphy_trees            = delphy.delphy_trees
+        File   delphy_log              = delphy.delphy_log
+        File   delphy_for_web_ui       = delphy.delphy_for_web_ui
+        File   delphy_output_for_beast = delphy.delphy_output_for_beast
+        File   delphy_stdout           = delphy.delphy_stdout
+        
+        String delphy_version          = delphy.delphy_version
+    }
+}
diff --git a/requirements-modules.txt b/requirements-modules.txt
index e0e4ab675..9221cd99c 100644
--- a/requirements-modules.txt
+++ b/requirements-modules.txt
@@ -5,6 +5,7 @@ broadinstitute/viral-phylo=2.1.20.2
 broadinstitute/py3-bio=0.1.2
 broadinstitute/beast-beagle-cuda=1.10.5pre
 broadinstitute/ncbi-tools=2.10.7.10
+broadinstitute/delphy=0.999
 nextstrain/base=build-20240318T173028Z
 andersenlabapps/ivar=1.3.1
 quay.io/staphb/pangolin=4.3.1-pdata-1.23.1