updates to some stat generation

Author: Jon Palmer

buscolite/__main__.py

@@ -26,11 +26,19 @@ def main():
         verbosity=3,
     )
     logger.info(
-        "Analysis complete:\n single-copy={}\n fragmented={}\n duplicated={}\n total={}".format(
+        "Assembly completeness:\n complete={:} [{:.2%}]\n single-copy={:} [{:.2%}]\n fragmented={:} [{:.2%}]\n duplicated={:} [{:.2%}]\n missing={:} [{:.2%}]\n total={:} [{:.2%}]".format(
+            stats["single-copy"] + stats["duplicated"],
+            ((stats["single-copy"] + stats["duplicated"]) / float(stats["total"])),
             stats["single-copy"],
+            (stats["single-copy"] / float(stats["total"])),
             stats["fragmented"],
+            (stats["fragmented"] / float(stats["total"])),
             stats["duplicated"],
+            (stats["duplicated"] / float(stats["total"])),
+            stats["missing"],
+            (stats["missing"] / float(stats["total"])),
             stats["total"],
+            (stats["total"] / float(stats["total"])),
         )
     )
     # write gff if genome mode

buscolite/__version__.py

@@ -1,3 +1,3 @@
-VERSION = (0, 5, 0)
+VERSION = (0, 5, 1)
 
 __version__ = ".".join(map(str, VERSION))

buscolite/busco.py

@@ -458,9 +458,14 @@ def runbusco(
         for b in CutOffs.keys():
             if b not in b_results:
                 missing.append(b)
-        stats = {"total": 0, "single-copy": 0, "fragmented": 0, "duplicated": 0}
+        stats = {
+            "total": len(CutOffs),
+            "single-copy": 0,
+            "fragmented": 0,
+            "duplicated": 0,
+            "missing": len(missing),
+        }
         for k, v in natsorted(b_results.items()):
-            stats["total"] += 1
             if len(v) > 1:  # duplicates
                 for i, x in enumerate(
                     sorted(v, key=lambda y: y["bitscore"], reverse=True)