Merge pull request #115 from broadinstitute/dp-nextstrain

add snp-sites task to nextstrain workflows
broadinstitute · Jun 11, 2020 · 1dc7d04 · 1dc7d04
2 parents 95d4550 + 80370a3
commit 1dc7d04
Show file tree

Hide file tree

Showing 4 changed files with 39 additions and 23 deletions.
diff --git a/pipes/WDL/tasks/tasks_nextstrain.wdl b/pipes/WDL/tasks/tasks_nextstrain.wdl
@@ -196,6 +196,30 @@ task augur_mafft_align {
     }
 }
 
+task snp_sites {
+    input {
+        File   msa_fasta
+        String docker = "quay.io/biocontainers/snp-sites:2.5.1--hed695b0_0"
+    }
+    String out_basename = basename(msa_fasta, ".fasta")
+    command {
+        snp-sites -V > VERSION
+        snp-sites -v -o ~{out_basename}.vcf ~{msa_fasta}
+    }
+    runtime {
+        docker: docker
+        memory: "1 GB"
+        cpu :   1
+        disks:  "local-disk 50 HDD"
+        preemptible: 0
+        dx_instance_type: "mem1_ssd1_v2_x2"
+    }
+    output {
+        File   snps_vcf = "~{out_basename}.vcf"
+        String snp_sites_version = read_string("VERSION")
+    }
+}
+
 task augur_mask_sites {
     meta {
         description: "Mask unwanted positions from alignment or SNP table. See https://nextstrain-augur.readthedocs.io/en/stable/usage/cli/mask.html"

diff --git a/pipes/WDL/workflows/augur_from_beast_mcc.wdl b/pipes/WDL/workflows/augur_from_beast_mcc.wdl
@@ -32,7 +32,6 @@ workflow augur_from_beast_mcc {
 
     output {
         File  beast_mcc_tree_newick      = augur_import_beast.tree_newick
-        File  node_data_json             = augur_import_beast.node_data_json
         File  auspice_input_json         = export_auspice_json.virus_json
     }
 }
diff --git a/pipes/WDL/workflows/augur_from_msa.wdl b/pipes/WDL/workflows/augur_from_msa.wdl
@@ -109,15 +109,9 @@ workflow augur_from_msa {
     }
 
     output {
-        File  masked_fasta               = augur_mask_sites.masked_sequences
-        File  raw_tree                   = draft_augur_tree.aligned_tree
-        File  refined_tree               = refine_augur_tree.tree_refined
-        File  branch_lengths             = refine_augur_tree.branch_lengths
-        File  json_nt_muts               = ancestral_tree.nt_muts_json
-        File  ancestral_sequences_fasta  = ancestral_tree.sequences
-        File  json_aa_muts               = translate_augur_tree.aa_muts_json
-        File? node_clade_data_json       = assign_clades_to_nodes.node_clade_data_json
-        File? json_ancestral_traits      = ancestral_traits.node_data_json
-        File  auspice_input_json         = export_auspice_json.virus_json
+        File  masked_alignment    = augur_mask_sites.masked_sequences
+        File  ml_tree             = draft_augur_tree.aligned_tree
+        File  time_tree           = refine_augur_tree.tree_refined
+        File  auspice_input_json  = export_auspice_json.virus_json
     }
 }
diff --git a/pipes/WDL/workflows/build_augur_tree.wdl b/pipes/WDL/workflows/build_augur_tree.wdl
@@ -64,6 +64,10 @@ workflow build_augur_tree {
             ref_fasta = ref_fasta,
             basename  = virus
     }
+    call nextstrain.snp_sites {
+        input:
+            msa_fasta = augur_mafft_align.aligned_sequences
+    }
     call nextstrain.augur_mask_sites {
         input:
             sequences = augur_mafft_align.aligned_sequences
@@ -125,17 +129,12 @@ workflow build_augur_tree {
     }
 
     output {
-        File  combined_assembly_fasta    = concatenate.combined
-        File  augur_aligned_fasta        = augur_mafft_align.aligned_sequences
-        File  masked_fasta               = augur_mask_sites.masked_sequences
-        File  raw_tree                   = draft_augur_tree.aligned_tree
-        File  refined_tree               = refine_augur_tree.tree_refined
-        File  branch_lengths             = refine_augur_tree.branch_lengths
-        File  json_nt_muts               = ancestral_tree.nt_muts_json
-        File  ancestral_sequences_fasta  = ancestral_tree.sequences
-        File  json_aa_muts               = translate_augur_tree.aa_muts_json
-        File? node_clade_data_json       = assign_clades_to_nodes.node_clade_data_json
-        File? json_ancestral_traits      = ancestral_traits.node_data_json
-        File  auspice_input_json         = export_auspice_json.virus_json
+        File  combined_assemblies = concatenate.combined
+        File  multiple_alignment  = augur_mafft_align.aligned_sequences
+        File  unmasked_snps       = snp_sites.snps_vcf
+        File  masked_alignment    = augur_mask_sites.masked_sequences
+        File  ml_tree             = draft_augur_tree.aligned_tree
+        File  time_tree           = refine_augur_tree.tree_refined
+        File  auspice_input_json  = export_auspice_json.virus_json
     }
 }