feat: added output of alignments to logging

3 files changed, 23 insertions, 7 deletions

diff --git a/Makefile b/Makefile
index b3f1d20..b0b9ffb 100644
--- a/Makefile
+++ b/Makefile
@@ -55,7 +55,7 @@ staph:
 	@echo "cluster    staph"; \
 	pangraph cluster -d data/staph data/staph/assemblies/*.fna.gz
 	@echo "build      staph"; \
-	pangraph build -d data/staph -m 500 -b 0 -e 2500 -w 1000 data/staph/guide.json 2>/dev/null 1>staph-e2500-w1000.log
+	pangraph build -d data/staph -m 500 -b 0 -e 2500 -w 1000 data/staph/guide.json 2>staph-e2500-w1000.err 1>staph-e2500-w1000.log
 
 # figures
 
diff --git a/pangraph/graph.py b/pangraph/graph.py
index c4e3f83..d154f74 100644
--- a/pangraph/graph.py
+++ b/pangraph/graph.py
@@ -518,12 +518,22 @@ class Graph(object):
                     tmp[1].flush()
 
                     def make_tree(n):
-                        proc = subprocess.Popen(f"mafft --auto {path[n]} | fasttree",
+                        proc = [None, None]
+                        out  = [None, None]
+                        err  = [None, None]
+                        proc[0] = subprocess.Popen(f"mafft --auto {path[n]}",
                                     stdout=subprocess.PIPE,
                                     stderr=subprocess.PIPE,
                                     shell=True)
-                        out, err = proc.communicate()
-                        tree = Phylo.read(io.StringIO(out.decode('utf-8')), format='newick')
+                        proc[1] = subprocess.Popen(f"fasttree",
+                                    stdin =subprocess.PIPE,
+                                    stdout=subprocess.PIPE,
+                                    stderr=subprocess.PIPE,
+                                    shell=True)
+                        out[0], err[0] = proc[0].communicate()
+                        out[1], err[1] = proc[1].communicate(input=out[0])
+                        tree = Phylo.read(io.StringIO(out[1].decode('utf-8')), format='newick')
+                        print(f"ALIGNMENT=\n{out[0]}")
                         print(f"SCORE={tree.total_branch_length()/(2*num_seqs)}", end=";")
 
                     make_tree(0)
diff --git a/scripts/parse_log.py b/scripts/parse_log.py
index bbca0ea..3a0aed0 100755
--- a/scripts/parse_log.py
+++ b/scripts/parse_log.py
@@ -12,10 +12,11 @@ score_preset = "SCORE="
 score_offset = len(score_preset)
 
 def main(args):
+    results = {}
     for log_path in args:
+        stats = defaultdict(lambda: {'hits':[], 'miss': 0})
+        level = -1
         with open(log_path) as log:
-            level = -1
-            stats = defaultdict(lambda: {'hits':[], 'miss': 0})
             for line in log:
                 line.rstrip('\n')
                 if line[0] == "+":
@@ -37,10 +38,15 @@ def main(args):
                         stats[level]['hits'].extend(score)
                         continue
                 raise ValueError(f"invalid syntax: {line[1:]}")
+        if len(stats) > 0:
+            path = log_path.replace(".log", "").split("-")
+            e, w = int(path[1][1:]), int(path[2][1:])
+            results[(e,w)] = dict(stats)
+    return results
 
 parser = argparse.ArgumentParser(description='process our data log files on end repair')
 parser.add_argument('files', type=str, nargs='+')
 
 if __name__ == "__main__":
     args = parser.parse_args()
-    main(args.files)
+    results = main(args.files)