2.6 draft 5

* Un-hide fungal species from NCBI and GBIF (this recovers about 2000 OTUs)
* Rewrite of uniquename generation logic, issue #38

Author: jar398

Makefile

@@ -5,7 +5,7 @@
 # Get it from http://files.opentreeoflife.org/ott/
 # and if there's a file "taxonomy" change that to "taxonomy.tsv".
 
-WHICH=2.6draft4
+WHICH=2.6draft5
 PREV_WHICH=2.5
 
 #  $^ = all prerequisites

make-ott.py

@@ -110,7 +110,7 @@
 # mark them 'hidden' so they can be suppressed downstream.  This
 # preserves the identifier assignments, which may have been used
 # somewhere.
-ncbi.taxon('Fungi').hideDescendants()
+ncbi.taxon('Fungi').hideDescendantsToRank('species')
 
 #ott.same(ncbi.taxon('Cyanobacteria'), silva.taxon('D88288/#3'))
 ott.notSame(ncbi.taxon('Burkea'), fung.taxon('Burkea'))
@@ -134,7 +134,7 @@
 gbif.smush()
 
 # Fungi suppressed at David Hibbett's request
-gbif.taxon('Fungi').hideDescendants()
+gbif.taxon('Fungi').hideDescendantsToRank('species')
 
 # Microbes suppressed at Laura Katz's request
 gbif.taxon('Bacteria','life').hideDescendants()
@@ -183,7 +183,7 @@
 irmng.smush()
 
 # Fungi suppressed at David Hibbett's request
-irmng.taxon('Fungi').hideDescendants()
+irmng.taxon('Fungi').hideDescendantsToRank('species')
 
 irmng.taxon('1413316').prune() #Neopithecus in Mammalia
 irmng.taxon('1413315').extinct() #Neopithecus in Primates (Pongidae)

org/opentreeoflife/smasher/Taxon.java

@@ -1003,20 +1003,47 @@ String uniqueName() {
 							  this.parent == null ? "(roots)" : this.parent.name);
 			return "?";
 		}
-		for (Taxon other : nodes)
-			if (other != this && other.name.equals(this.name)) {
+		boolean homonymp = false;
+		boolean informativeFail = false;
+
+		if (true) {
+			for (Taxon other : nodes)
+				if (other != this && other.name.equals(this.name)) {
+					homonymp = true;
+					Taxon i = this.informative();
+					if (i != null && i.equals(other.informative())) {
+						informativeFail = true;
+						break;
+					}
+				}
+			if (informativeFail || homonymp) {
+				String urank = "";
+				if (this.rank != null) urank = this.rank + " ";
+				if (informativeFail)
+					urank = urank + " " + this.sourceIds.get(0);
+
 				Taxon i = this.informative();
-				if ((i != other.informative() &&
-					 i != null &&
-					 !this.name.endsWith(" sp."))) {
-					String urank = "";
-					if (this.rank != null) urank = this.rank + " ";
-					String irank = "";
-					if (i.rank != null) irank = i.rank + " ";
-					return this.name + " (" + urank + "in " + irank + i.name + ")";
-				} else
-					return this.name + " (" + this.getSourceIdsString() + ")";
-			}
+				String irank = "";
+				if (i.rank != null) irank = i.rank + " ";
+				return this.name + " (" + urank + "in " + irank + i.name + ")";
+			} else return "";
+		}
+		else
+			// Old buggy version, delete after above has been tested
+			for (Taxon other : nodes)
+				if (other != this && other.name.equals(this.name)) {
+					Taxon i = this.informative();
+					if ((i != other.informative() &&
+						 i != null &&
+						 !this.name.endsWith(" sp."))) {
+						String urank = "";
+						if (this.rank != null) urank = this.rank + " ";
+						String irank = "";
+						if (i.rank != null) irank = i.rank + " ";
+						return this.name + " (" + urank + "in " + irank + i.name + ")";
+					} else
+						return this.name + " (" + this.getSourceIdsString() + ")";
+				}
 		return "";
 	}
 
@@ -1101,6 +1128,20 @@ Taxon sample(int k, Taxonomy tax) {
 		return sam;
 	}
 
+	public boolean isHidden() {
+		return ((this.properFlags | this.inheritedFlags) &
+				(Taxonomy.HIDDEN |
+				 Taxonomy.EXTINCT |
+				 Taxonomy.MAJOR_RANK_CONFLICT |
+				 Taxonomy.TATTERED |
+				 Taxonomy.NOT_OTU |
+				 Taxonomy.HYBRID |
+				 Taxonomy.VIRAL |
+				 Taxonomy.UNCLASSIFIED |
+				 Taxonomy.ENVIRONMENTAL |
+				 Taxonomy.INCERTAE_SEDIS)) != 0;
+	}
+
 	// ----- Methods intended for use in jython scripts -----
 
 	// Patch system commands are add, move, synonym, prune, fold, flag
@@ -1157,6 +1198,17 @@ public void hideDescendants() {
 			}
 	}
 
+	// Hide up to but not including the given rank
+	public void hideDescendantsToRank(String rank) {
+		if (this.children != null)
+			for (Taxon child : this.children) {
+				if (child.rank != null && !child.rank.equals(rank)) {
+					child.properFlags = Taxonomy.HIDDEN;
+					child.hideDescendantsToRank(rank);
+				}
+			}
+	}
+
 	public void synonym(String name) {
 		this.taxonomy.addSynonym(name, this);
 	}
@@ -1176,7 +1228,7 @@ public void elide() {
 			System.err.format("** %s is a root\n", this);
 		else {
 			if (this.children == null)
-				System.err.format("** Warning: %s has no children\n", this);
+				; //System.err.format("** Warning: %s has no children\n", this);
 			else
 				for (Taxon child : new ArrayList<Taxon>(this.children))
 					child.changeParent(this.parent);

org/opentreeoflife/smasher/Taxonomy.java

@@ -725,18 +725,8 @@ void dumpSynonyms(String filename, String sep) throws IOException {
 	void dumpHidden(String filename) throws IOException {
 		PrintStream out = Taxonomy.openw(filename);
 		for (Taxon node : this) {
-			if (((node.properFlags | node.inheritedFlags) &
-				 (Taxonomy.HIDDEN |
-				  Taxonomy.EXTINCT |
-				  Taxonomy.MAJOR_RANK_CONFLICT |
-				  Taxonomy.TATTERED |
-				  Taxonomy.NOT_OTU |
-				  Taxonomy.HYBRID |
-				  Taxonomy.VIRAL |
-				  Taxonomy.UNCLASSIFIED |
-				  Taxonomy.ENVIRONMENTAL |
-				  Taxonomy.INCERTAE_SEDIS)) != 0) {
-				out.format("%s\t%s\t%s\t", node.id, node.name, node.division);
+			if (node.isHidden()) {
+				out.format("%s\t%s\t%s\t%s\t", node.id, node.name, node.getSourceIdsString(), node.division);
 				this.printFlags(node, out);
 				out.println();
 			}
@@ -966,7 +956,7 @@ static void analyzeContainers(Taxon node, int inheritedFlags) {
 		boolean elidep = false;
 
 		if (unclassifiedRegex.matcher(node.name).find()) {// Rule 3+5
-			node.properFlags |= UNCLASSIFIED;
+			node.properFlags |= UNCLASSIFIED;  // includes uncultured
 			elidep = true;
 		}
 		if (environmentalRegex.matcher(node.name).find()) {// Rule 3+5