pruning large

README

@@ -1,4 +1,4 @@
-uDance version 1.2.1
+uDance version 1.3.0
 
 AUTHORS:
 Metin Balaban
@@ -10,6 +10,10 @@ Anaconda
 
 Changelog:
 
+1.3.0:
+-Pruning Large partitions
+-Changes in TreeCluster logic disallowing formation of very small partitions
+
 1.2.1:
 -APPLES2 excludes sequences that are placed on internal nodes.
 -Filtered backbone sequences are no longer added to the query set.

config.yaml

@@ -21,9 +21,9 @@ mainlines_config:
 apples_config:
   # [FM, OLS]
   method: FM
-  # float. (0-infinity)
+  # float. [0-infinity)
   filter: 0.2
-  # integer. (5-infinity)
+  # integer. [5-infinity)
   base: 25
   # minimum overlap fraction. Default  is 0.001
   overlap: 0.05
@@ -37,6 +37,10 @@ prep_config:
   sublength: 100
   # minimum fragment length
   fraglength: 75
+  # large clusters are pruned. ASTRAL's limit is around 9000 for 400+ genes.
+  pruneafter: 9000
+  # [0-1]. lower threshold = stricter pruning
+  prune_thr: 0.96
 
 infer_config:
   # [raxml-ng,iqtree,raxml-8]

prune_similar.py

@@ -15,9 +15,9 @@
     parser.add_option("-T", "--threads", type=int, dest="num_thread", default=0,
                       help="number of cores used in placement. "
                            "0 to use all cores in the running machine", metavar="NUMBER")
-    parser.add_option("-c", "--cutoff", type=float, dest="cutoff_threshold", metavar='NUMBER', default=0.95,
+    parser.add_option("-c", "--cutoff", type=float, dest="cutoff_threshold", metavar='NUMBER', default=0.96,
                       help="threshold number of dissimilar genes.")
-    parser.add_option("-S", "--size", type=int, dest="minimum_size", metavar='NUMBER', default=8000,
+    parser.add_option("-S", "--size", type=int, dest="minimum_size", metavar='NUMBER', default=9000,
                       help="partition size requirement for pruning.")
 
     (options, args) = parser.parse_args()

uDance/subsample_partition.py

@@ -95,6 +95,13 @@ def subsample_partition(partition_output_dir, cutoff):
         return ""
     pruned_species = set(pruned_species)
 
+    # remove pruned from species.txt
+    with open(join(partition_output_dir, "species.txt")) as f:
+        prevspecies = set(map(lambda x: x.strip(), f.readlines()))
+    newspecieslst = list(prevspecies.difference(pruned_species))
+    with open(join(partition_output_dir, "species.txt"), "w") as f:
+        f.write("\n".join(pruned_species) + "\n")
+
     for g in genes:
         aln_dict = dict()
         with open(join(g, "aln.fa")) as af:
@@ -131,12 +138,12 @@ def subsample_partition(partition_output_dir, cutoff):
                 if len(v) > 1:
                     duplist.append("\t".join(v))
 
-            aln_output_path = join(g, "aln_pruned.fa")
+            aln_output_path = join(g, "aln.fa")
             with open(aln_output_path, "w", buffering=100000000) as f:
                 f.write("\n".join(res))
                 f.write("\n")
             if duplist:
-                dupmap_output_path = join(g, "dupmap_pruned.txt")
+                dupmap_output_path = join(g, "dupmap.txt")
                 with open(dupmap_output_path, "w", buffering=100000000) as f:
                     f.write("\n".join(duplist))
                     f.write("\n")

uDance/treecluster_sum.py

@@ -71,6 +71,7 @@ def min_tree_coloring_sum_max(tree, thr, max_thr):
         else:
             left, right = current.children
             if left.weight + right.weight + current.weight <= thr or \
+                    (left.weight + right.weight + current.weight > thr and left.weight + right.weight <= max(3, thr / 10)) or \
                     left.edge_length + left.farthest + right.edge_length + right.farthest < max_thr or \
                     (left.edge_length <= ZERO_LEN and len(left.placements) > 0) or \
                     (right.edge_length <= ZERO_LEN and len(right.placements) > 0) or \

udance.smk

@@ -197,6 +197,8 @@ checkpoint decompose:
         method=config["infer_config"]["method"],
         sub=config["prep_config"]["sublength"],
         frag=config["prep_config"]["fraglength"],
+        pra=config["prep_config"]["pruneafter"],
+        prt=config["prep_config"]["prune_thr"],
         char=config["chartype"]
 
     resources: cpus=config["resources"]["cores"],
@@ -210,7 +212,7 @@ checkpoint decompose:
             else
                 python run_udance.py decompose -p -s {input.ind} -o {outdir}/udance -t {params.size} -j {input.j} -m {params.method} -T {resources.cpus} -l {params.sub} -f {params.frag}
             fi
-            python prune_similar.py -T {resources.cpus} -o {outdir}/udance
+            python prune_similar.py -T {resources.cpus} -o {outdir}/udance -S {params.pra} -c {params.prt}
             if [  -f {outdir}/udance/dedupe_map.txt ]; then 
                 cat {outdir}/udance/dedupe_map.txt > {outdir}/dedupe_map.txt 
             fi