From 001c1dcac1aa901dbc4aad4225617857e1b682b8 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Sun, 9 Oct 2016 04:24:08 -0300
Subject: [PATCH] For a given protein consider only cluster with at list a
 score of 95% of the best score

Former-commit-id: cfdc6fcd37a4036d8bcca27bc7e120e60a94998d
Former-commit-id: f45bb7922f28165fd3baa1bc67bf815a759d1590
---
 detectors/cds/lib/bestclust.awk | 23 ++++++++++++++++++++---
 1 file changed, 20 insertions(+), 3 deletions(-)

diff --git a/detectors/cds/lib/bestclust.awk b/detectors/cds/lib/bestclust.awk
index 5354a0a..5634516 100644
--- a/detectors/cds/lib/bestclust.awk
+++ b/detectors/cds/lib/bestclust.awk
@@ -77,7 +77,7 @@ BEGIN {
 }
 
 END {
-
+  ClustScoreMax=0
   # make clusters
   pi = -1
   for (i in Cover) {
@@ -103,12 +103,29 @@ END {
       Clust[clusno]["score"] = score
       Clust[clusno]["strand"] = Entry[i]["strand"]
       Clust[clusno]["entry"] = i
+      
+      if (score > ClustScoreMax)
+      	ClustScoreMax=score
     }
   }
 
-  # print cluster info
-  print "c nclust", NbClust+0
+  
+  # Consider only cluster with at least 95% of the best score
+  
+  NbClustOk=0
   for (i = 1 ; i <= NbClust ; i++) {
+    if (Clust[i]["score"] >= ClustScoreMax * 0.95) 
+    	NbClustOk++
+    else
+    	Clust[i]["score"]=0
+
+  }
+
+
+  # print cluster info
+  print "c nclust", NbClustOk+0
+  for (i = 1 ; i <= NbClust ; i++) {
+  	if (Clust[i]["score"] == 0) continue
     print "c cluster", i, "from", Clust[i]["from"], "to", Clust[i]["to"],\
           "strand", Clust[i]["strand"], "score", Clust[i]["score"]
   }