From d2aabe1ef9c72d628e5945032ce70b0ece546db4 Mon Sep 17 00:00:00 2001
From: Robert Bossy <Robert.Bossy@inra.fr>
Date: Fri, 24 Sep 2021 11:26:32 +0200
Subject: [PATCH] reject ambiguous name categories

---
 rewrite-taxonomy.snakefile | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/rewrite-taxonomy.snakefile b/rewrite-taxonomy.snakefile
index 77b4b3f..87cbf43 100644
--- a/rewrite-taxonomy.snakefile
+++ b/rewrite-taxonomy.snakefile
@@ -44,7 +44,7 @@ rule taxaid_full:
         config['OUTDIR'] + '/' + config['DSMZ_MATCH_DIR']
 
     shell:
-        '''{config[REWRITE_TAXONOMY]} -namesFile {config[OUTDIR]}/{config[NCBI_DIR]}/names.dmp -namesFile {input}/dsmz-names.dmp -prefix ncbi: -rejectionFile reject.txt -saturationFile saturate.txt -pattern '{{NAME}}\t{{TAXID}}\t{{CANONICAL}}\t{{TAXID_PATH}}\t{{POS_TAG}}\t{{RANK}}\t{{SPECIES_TAXID}}\t{{SPECIES_NAME}}\n' {config[OUTDIR]}/{config[NCBI_DIR]}/nodes.dmp {input}/dsmz-nodes.dmp >{output}'''
+        '''{config[REWRITE_TAXONOMY]} -namesFile {config[OUTDIR]}/{config[NCBI_DIR]}/names.dmp -namesFile {input}/dsmz-names.dmp -prefix ncbi: -rejectionFile reject.txt -rejectNameType in-part -rejectNameType Includes -saturationFile saturate.txt -pattern '{{NAME}}\t{{TAXID}}\t{{CANONICAL}}\t{{TAXID_PATH}}\t{{POS_TAG}}\t{{RANK}}\t{{SPECIES_TAXID}}\t{{SPECIES_NAME}}\n' {config[OUTDIR]}/{config[NCBI_DIR]}/nodes.dmp {input}/dsmz-nodes.dmp >{output}'''
 
 
 rule taxid_full:
@@ -55,4 +55,4 @@ rule taxid_full:
         config['OUTDIR'] + '/' + config['DSMZ_MATCH_DIR']
 
     shell:
-        '''{config[REWRITE_TAXONOMY]} -taxaDict -namesFile {config[OUTDIR]}/{config[NCBI_DIR]}/names.dmp -namesFile {input}/dsmz-names.dmp  -prefix ncbi: -rejectionFile reject.txt -saturationFile saturate.txt -pattern '{{TAXID}}\t{{CANONICAL}}\t{{TAXID_PATH}}\t{{RANK}}\n' {config[OUTDIR]}/{config[NCBI_DIR]}/nodes.dmp {input}/dsmz-nodes.dmp >{output}'''
+        '''{config[REWRITE_TAXONOMY]} -taxaDict -namesFile {config[OUTDIR]}/{config[NCBI_DIR]}/names.dmp -namesFile {input}/dsmz-names.dmp  -prefix ncbi: -rejectionFile reject.txt -rejectNameType in-part -rejectNameType Includes -saturationFile saturate.txt -pattern '{{TAXID}}\t{{CANONICAL}}\t{{TAXID_PATH}}\t{{RANK}}\n' {config[OUTDIR]}/{config[NCBI_DIR]}/nodes.dmp {input}/dsmz-nodes.dmp >{output}'''
-- 
GitLab