Added files from NCBI taxonomy db.

updated script to get the files

Signed-off-by: Inti Pedroso <[email protected]>

Author: inti

example/run.sh

@@ -2,12 +2,12 @@
 
 # run test of script.
 
-my_perl phylostratiphy.pl \
-    -blast blastp__Aech_v3.8.pep__nr.e10.sw.txt_old \
-    -tax_folder data/ \
+my_perl ../phylostratiphy.pl \
+    -blast example/blastp__Aech_v3.8.pep__nr.e10.sw.txt_old \
+    -tax_folder ncbi_tax_data/ \
     -prot_only \
     -query_taxon 103372 \
-    -virus_list tmp.virus.txt \
+    -virus_list example/virus_list.tx \
     -out test_phylostratiphy
 
 

example/run2.sh

@@ -2,13 +2,16 @@
 
 # run test of script.
 
+cd ..
+
 perl phylostratiphy.pl \
-    -blast dysbindin.blast_out.txt \
-    -tax_folder data/ \
+    -blast example/dysbindin.blast_out.txt \
+    -tax_folder ncbi_tax_data/ \
     -query_taxon 9606 \
-	 -gi_tax_id dysbindin.tax_info.csv  \
+	 -gi_tax_id example/dysbindin.tax_info.csv  \
     -hard 1e-3 \
-    -virus_list virus_list.txt \
-    -out test_phylostratiphy
+    -virus_list example/virus_list.txt \
+    -out example/test_phylostratiphy
 
+cd example
 

get_taxonomy_data.sh

@@ -1,5 +1,5 @@
 echo Starting to download NCBI taxonomy db info
-out_folder=data
+out_folder=ncbi_tax_data
 
 mkdir $out_folder
 
@@ -9,14 +9,21 @@ cd $out_folder
 tax_ftp="ftp://ftp.ncbi.nih.gov/pub/taxonomy"
 echo Data from FTP folder [ $tax_ftp ]
 
-for file in gi_taxid_prot.dmp.gz gi_taxid_nucl.dmp.gz taxcat.tar.gz taxdump.tar.gz taxdump_readme.txt gi_taxid.readme taxcat_readme.txt 
+for file in taxdump.tar.gz taxdump_readme.txt
 do 
 	wget $tax_ftp/$file
 done
-
+			
 echo Uncompressing
 ls *.gz | xargs gunzip -v
 ls *.tar | perl -ne 'chomp($_); system "tar xvf $_";'
+			
+rm citations.dmp delnodes.dmp division.dmp gc.prt gencode.dmp merged.dmp taxdump.tar
+
+echo Files updated on [ `date` ] > files_update_info.txt
+
+cd ..
 
 echo Done  
 
+

ncbi_tax_data/files_update_info.txt

@@ -0,0 +1 @@
+Files updated on [ Wed 25 Jul 2012 16:18:05 BST ]