Fix issues in processing contaminants refseq genomes, including do no…

…t chopping off for the pipe sign in the header

centrifuger-download

@@ -13,8 +13,13 @@ cut_after_first_space_or_second_pipe() {
 }
 export -f cut_after_first_space_or_second_pipe
 
+cut_after_first_space() {
+    grep '^>' | sed 's/ .*//' 
+}
+export -f cut_after_first_space
+
 map_headers_to_taxid() {
-    grep '^>' | cut_after_first_space_or_second_pipe | sed -e "s/^>//" -e  "s/\$/    $1/"
+    grep '^>' | cut_after_first_space | sed -e "s/^>//" -e  "s/\$/    $1/"
 }
 export -f map_headers_to_taxid
 

indices/Makefile

@@ -285,7 +285,8 @@ endif
 $(REFERENCE_SEQUENCES_DIR)/contaminants.fna: | $(REFERENCE_SEQUENCES_DIR)
 	[[ -d $(TMP_DIR) ]] && rm -rf $(TMP_DIR); mkdir -p $(TMP_DIR)
 	centrifuger-download -o $(TMP_DIR) contaminants > $(TMP_DIR)/$(patsubst %.fna,%$(TAXID_SUFFIX),$(notdir $@))
-	find $(TMP_DIR)/contaminants -name "*.fna.gz" | xargs zcat > $@.tmp && mv $@.tmp $@
+	#find $(TMP_DIR)/contaminants -name "*.fna.gz" | xargs zcat > [email protected] && mv [email protected] $@
+	find $(TMP_DIR)/contaminants -name "*.fna" | xargs cat > $@.tmp && mv $@.tmp $@
 	mv $(TMP_DIR)/$(patsubst %.fna,%$(TAXID_SUFFIX),$(notdir $@)) $(patsubst %.fna,%$(TAXID_SUFFIX),$@)
 ifeq (1,$(KEEP_FILES))
 	[[ -d $(DL_DIR)/contaminants ]] || mkdir -p $(DL_DIR)/contaminants