Merge pull request #370 from nextstrain/export_auth

Reduce No Author Errors
nextstrain · Sep 5, 2019 · 3435007 · 3435007
2 parents f419898 + 7e953e2
commit 3435007
Show file tree

Hide file tree

Showing 4 changed files with 244 additions and 1 deletion.
diff --git a/augur/export.py b/augur/export.py
@@ -272,9 +272,10 @@ def construct_author_info_nexflu(metadata, tree, nodes):
             authorsInTree.add(nodes[node.name]["authors"])
 
     author_info = defaultdict(lambda: {"n": 0})
+    no_authors = 0
     for strain, data in metadata.items():
         if "authors" not in data:
-            print("Error - {} had no authors".format(strain))
+            no_authors += 1
             continue
         if data["authors"] not in authorsInTree:
             continue
@@ -286,6 +287,8 @@ def construct_author_info_nexflu(metadata, tree, nodes):
                 if attr in author_info[authors] and data[attr].strip() != author_info[authors][attr].strip():
                     print("Error - {} had contradictory {}(s): {} vs {}".format(authors, attr, data[attr], author_info[authors][attr]))
                 author_info[authors][attr] = data[attr].strip()
+    if no_authors:
+        print("Warning - {} samples did not have author information.".format(no_authors))
 
     return author_info
 

diff --git a/tests/builds/tb/Snakefile b/tests/builds/tb/Snakefile
@@ -2,6 +2,8 @@ rule all:
     input:
         auspice_tree = "auspice/tb_tree.json",
         auspice_meta = "auspice/tb_meta.json",
+        auspice_noauth_tree = "auspice/tb-noauth_tree.json",
+        auspice_noauth_meta = "auspice/tb-noauth_meta.json",
         auspice_seq = "auspice/tb_seq.json",
         auspice_seq_v2 = "auspice/tb_seq_v2.json",
         auspice_main = "auspice/tb.json"
@@ -13,6 +15,7 @@ rule config:
     params:
         seq = "data/tb.vcf.gz",
         meta = "data/meta.tsv",
+        no_auth_meta = "data/meta-noAuth.tsv",
         exclude = "data/dropped_strains.txt",
         mask = "data/Locus_to_exclude_Mtb.bed",
         ref = "data/ref.fasta",
@@ -22,6 +25,7 @@ rule config:
         genes = "data/genes.txt",
         colors = "data/color.tsv",
         config = "data/config.json",
+        no_auth_config = "data/no-auth-config.json",
         geo_info = "data/lat_longs.tsv",
         clades = "data/clades.tsv"
 
@@ -209,6 +213,38 @@ rule export:
         augur validate --json {output.tree} {output.meta}
         """
 
+rule no_auth_export:
+    message: "Exporting data files for for auspice using nextflu compatible schema"
+    input:
+        tree = rules.refine.output.tree,
+        metadata = config.no_auth_meta,
+        branch_lengths = rules.refine.output.node_data,
+        traits = rules.traits.output,
+        nt_muts = rules.ancestral.output.nt_data,
+        aa_muts = rules.translate.output.aa_data,
+        color_defs = config.colors,
+        config = config.no_auth_config,
+        geo_info = config.geo_info,
+        clades = rules.clades.output.clade_data,
+        ref = config.ref,
+        translations = rules.translate.output.fasta_out
+    output:
+        tree = rules.all.input.auspice_noauth_tree,
+        meta = rules.all.input.auspice_noauth_meta
+    shell:
+        """
+        augur export \
+            --tree {input.tree} \
+            --metadata {input.metadata} \
+            --reference {input.ref} --reference-translations {input.translations} \
+            --node-data {input.branch_lengths} {input.traits} {input.aa_muts} {input.nt_muts} {input.clades} \
+            --auspice-config {input.config} \
+            --colors {input.color_defs} \
+            --lat-longs {input.geo_info} \
+            --output-tree {output.tree} \
+            --output-meta {output.meta} 
+        augur validate --json {output.tree} {output.meta}
+        """
 
 rule exportv2:
     message: "Exporting data files for for auspice using nextstrain schema v2"