Use Nextclade3

nextstrain · Apr 30, 2024 · 27fdc7b · 27fdc7b
1 parent b0d1ed4
commit 27fdc7b
Show file tree

Hide file tree

Showing 8 changed files with 415 additions and 22 deletions.
diff --git a/phylogenetic/build-configs/ci/config.yaml b/phylogenetic/build-configs/ci/config.yaml
@@ -2,7 +2,7 @@ custom_rules:
   - build-configs/ci/copy_example_data.smk
 
 reference: "defaults/reference.fasta"
-genemap: "defaults/genemap.gff"
+genome_annotation: "defaults/genome_annotation.gff3"
 genbank_reference: "defaults/reference.gb"
 include: "defaults/hmpxv1/include.txt"
 clades: "defaults/clades.tsv"

diff --git a/phylogenetic/defaults/exclude_accessions.txt b/phylogenetic/defaults/exclude_accessions.txt
@@ -75,7 +75,4 @@ PP098578
 
 HM172544	# cidofovir-resistant lab strain that is derived from DQ011155 (h/t Andrew Rambaut)
 
-TMP0003	# Overdiverged 23MPX1786C
-TMP0045	# Overdiverged RDC-NKV-GOM-MPOX-004
-
 NC_003310	# Overdiverged RefSeq NC_003310
diff --git a/phylogenetic/defaults/genome_annotation.gff3 b/phylogenetic/defaults/genome_annotation.gff3
diff --git a/phylogenetic/defaults/hmpxv1/config.yaml b/phylogenetic/defaults/hmpxv1/config.yaml
@@ -1,5 +1,5 @@
 reference: "defaults/reference.fasta"
-genemap: "defaults/genemap.gff"
+genome_annotation: "defaults/genome_annotation.gff3"
 genbank_reference: "defaults/reference.gb"
 include: "defaults/hmpxv1/include.txt"
 clades: "defaults/clades.tsv"

diff --git a/phylogenetic/defaults/hmpxv1_big/config.yaml b/phylogenetic/defaults/hmpxv1_big/config.yaml
@@ -1,5 +1,5 @@
 reference: "defaults/reference.fasta"
-genemap: "defaults/genemap.gff"
+genome_annotation: "defaults/genome_annotation.gff3"
 genbank_reference: "defaults/reference.gb"
 include: "defaults/hmpxv1_big/include.txt"
 clades: "defaults/clades.tsv"

diff --git a/phylogenetic/defaults/mpxv/config.yaml b/phylogenetic/defaults/mpxv/config.yaml
@@ -1,7 +1,7 @@
 auspice_config: "defaults/mpxv/auspice_config.json"
 include: "defaults/mpxv/include.txt"
 reference: "defaults/reference.fasta"
-genemap: "defaults/genemap.gff"
+genome_annotation: "defaults/genome_annotation.gff3"
 genbank_reference: "defaults/reference.gb"
 lat_longs: "defaults/lat_longs.tsv"
 description: "defaults/description.md"

diff --git a/phylogenetic/rules/annotate_phylogeny.smk b/phylogenetic/rules/annotate_phylogeny.smk
@@ -48,15 +48,15 @@ rule translate:
     input:
         tree=build_dir + "/{build_name}/tree.nwk",
         node_data=build_dir + "/{build_name}/nt_muts.json",
-        genemap=config["genemap"],
+        genome_annotation=config["genome_annotation"],
     output:
         node_data=build_dir + "/{build_name}/aa_muts.json",
     shell:
         """
         augur translate \
             --tree {input.tree} \
             --ancestral-sequences {input.node_data} \
-            --reference-sequence {input.genemap} \
+            --reference-sequence {input.genome_annotation} \
             --output {output.node_data}
         """
 

diff --git a/phylogenetic/rules/prepare_sequences.smk b/phylogenetic/rules/prepare_sequences.smk
@@ -5,7 +5,7 @@ REQUIRED INPUTS:
 
     include     = path to file of sequences to in force include
     reference   = path to reference sequence FASTA for Nextclade alignment
-    genemap     = path to genemap GFF for Nextclade alignment
+    genome_annotation     = path to genome_annotation GFF for Nextclade alignment
     maskfile    = path to maskfile of sites to be masked
 
 OUTPUTS:
@@ -158,30 +158,35 @@ rule reverse_reverse_complements:
 rule align:
     """
     Aligning sequences to {input.reference}
-      - filling gaps with N
     """
     input:
         sequences=build_dir + "/{build_name}/reversed.fasta",
         reference=config["reference"],
-        genemap=config["genemap"],
+        genome_annotation=config["genome_annotation"],
     output:
         alignment=build_dir + "/{build_name}/aligned.fasta",
-        insertions=build_dir + "/{build_name}/insertions.fasta",
     params:
-        max_indel=config["max_indel"],
-        seed_spacing=config["seed_spacing"],
+        # Alignment params from all-clades nextclade dataset
+        excess_bandwidth=100,
+        terminal_bandwidth=300,
+        window_size=40,
+        min_seed_cover=0.1,
+        allowed_mismatches=8,
+        gap_alignment_side="left",
     threads: workflow.cores
     shell:
         """
-        nextalign run \
+        nextclade3 run \
             --jobs {threads} \
-            --reference {input.reference} \
-            --genemap {input.genemap} \
-            --max-indel {params.max_indel} \
-            --seed-spacing {params.seed_spacing} \
-            --retry-reverse-complement \
+            --input-ref {input.reference} \
+            --input-annotation {input.genome_annotation} \
+            --excess-bandwidth {params.excess_bandwidth} \
+            --terminal-bandwidth {params.terminal_bandwidth} \
+            --window-size {params.window_size} \
+            --min-seed-cover {params.min_seed_cover} \
+            --allowed-mismatches {params.allowed_mismatches} \
+            --gap-alignment-side {params.gap_alignment_side} \
             --output-fasta - \
-            --output-insertions {output.insertions} \
             {input.sequences} | seqkit seq -i > {output.alignment}
         """