From b089e836608dfeefe89969ddcd7fe60bd47ef7e2 Mon Sep 17 00:00:00 2001
From: Pjotr Prins <pjotr.public01@thebird.nl>
Date: Sat, 26 Jul 2014 20:35:09 +0200
Subject: [PATCH] table

---
 sparql/extra/gwp/blast2.rq     |  2 +-
 sparql/extra/gwp/gen_table.sh  |  2 +-
 sparql/extra/gwp/gen_table2.rb | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/sparql/extra/gwp/blast2.rq b/sparql/extra/gwp/blast2.rq
index 6ecc8b4..4a9b2fb 100644
--- a/sparql/extra/gwp/blast2.rq
+++ b/sparql/extra/gwp/blast2.rq
@@ -28,7 +28,7 @@
 <% end %>
 {
 
-# :Mi_CDS_cluster00100_NP_001022867 :cluster :Mi_CDS_cluster00100 .
+# :Mi_CDS_cluster00100_NP_001022867 gwp:cluster :Mi_CDS_cluster00100 .
 # :Mi_CDS_cluster00100_NP_001022867 a :blast_match .
 # :Mi_CDS_cluster00100_NP_001022867 rdf:label "Mi_CDS_cluster00100_NP_001022867" .
 # :Mi_CDS_cluster00100_NP_001022867 :homolog_species "Ce" .
diff --git a/sparql/extra/gwp/gen_table.sh b/sparql/extra/gwp/gen_table.sh
index dd9494c..f3b9733 100755
--- a/sparql/extra/gwp/gen_table.sh
+++ b/sparql/extra/gwp/gen_table.sh
@@ -14,4 +14,4 @@ for species in Bm Bx Ce Cb Gp Mi Mh Pi Pp Sr Ts ; do
     ./gen_table2.rb species=$species source=$source |egrep ^cat >> $fn
   done
 done
-bio-table --merge tmp/* > table2.tsv
+bio-table --merge tmp/*.tsv > table2.tsv
diff --git a/sparql/extra/gwp/gen_table2.rb b/sparql/extra/gwp/gen_table2.rb
index 35db21e..a171f8b 100755
--- a/sparql/extra/gwp/gen_table2.rb
+++ b/sparql/extra/gwp/gen_table2.rb
@@ -52,7 +52,7 @@
 # Note that catH overlaps with catA and that catH is larger than all PSC(!)
 catH = csv_parse.call("env HASH=\"by_cluster=1,species=#{species},source1=#{TYPE}\" ../../../scripts/sparql-csv.sh blast2.rq").drop(1).flatten
 # ==== catH
-assert(catH.size == 29,"Expect 29 was #{catH.size}") if do_assert
+assert(catH.size == 36,"Expect 36 was #{catH.size}") if do_assert
 
 # ---- 2b. Annotate plantP only (&)
 #      catH contains all ann PSC. So we can select those that
@@ -123,7 +123,7 @@
 
 minc_cluster_plantp = ann.keys.select { |k| ann[k].include?(:plant_pathogen) }
 p [:plantP, minc_cluster_plantp.size ]
-assert(minc_cluster_plantp.size == 13) if do_assert 
+assert(minc_cluster_plantp.size == 9) if do_assert 
 
 # ---- 2c and 2d. Annotated in ann! (&)
 
@@ -140,16 +140,16 @@
 # ==== we have catA orange and red!
 otherA = catA - plant_pathogenA
 p [:orange, otherA.size]
-assert(otherA.size == 4,otherA.size.to_s) if do_assert 
+assert(otherA.size == 6,otherA.size.to_s) if do_assert 
 
 # ---- Now we can have the unique PSC (catC green) (&)
 catC = all - catA - catH
 p [:unique_PSC, catC.size]
-assert(catC.size == 14,"Was #{catC.size}") if do_assert
+assert(catC.size == 7,"Was #{catC.size}") if do_assert
 
 assert(catA & catC == [],"There should be no overlap between catA and catC")
 catB = all - catA - catC 
-assert(catB.size == 21,catB.size) if do_assert
+assert(catB.size == 28,catB.size) if do_assert
 
 # ---- Fetch conserved (catB)
 p '** all **********************'