addressing #63

zhangrengang · Dec 25, 2024 · 0072927 · 0072927
1 parent 36793dd
commit 0072927
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 3 deletions.
diff --git a/TEsorter/app.py b/TEsorter/app.py
@@ -496,9 +496,10 @@ def identify_rexdb(self, genes, clades):
 			('LTR', 'Bel-Pao'): ['GAG', 'PROT', 'RT', 'RH', 'INT'],
 			}
 		clade_count = Counter(clades)
+		counts = list(clade_count.values())
 		max_clade = max(clade_count, key=lambda x: clade_count[x])
 		order, superfamily = self._parse_rexdb(max_clade)
-		if len(clade_count) == 1 or clade_count[max_clade] > 1:
+		if len(clade_count) == 1 or (clade_count[max_clade] > 1 and counts[0] > counts[1]):
 			max_clade = max_clade.split('/')[-1]
 		elif len(clade_count) > 1:
 			max_clade = 'mixture'

diff --git a/scripts/LTR_tree.R b/scripts/LTR_tree.R
@@ -21,6 +21,7 @@ format_id <- function(x1, x2, x3, x4) {
 	x1 = sapply(x1, split_id)	
 	x1 = gsub('\\W+', '_', x1)
 	x = paste(x1, x2, x3, x4, sep='_')
+	x = gsub('\\W+', '_', x)
 	return(x)
 }
 
@@ -35,8 +36,10 @@ if (branch_color == 'Clade') {
 			labels = map[which(map$Clade==clade), ]
 			clade = paste(labels$Superfamily, labels$Clade, sep='/')[1]
 			labels = format_id(labels$X.TE, labels$Order, labels$Superfamily, labels$Clade)
-			if (! any(labels %in% tree$tip.label)) {next}
-	#		print(clade)
+			if (! any(labels %in% tree$tip.label)) {
+				#print(labels); 
+				next}
+			#print(clade)
 			grp[[clade]] = labels
 	}
 	clades = sort(names(grp))