Add summary for excluded matches

fixcontigs.py

@@ -339,6 +339,25 @@ def parse_valid_matches(matches, out_prefix, reference_fasta):
           (GAP_BLOCK_SIZE_MIN, n_prot_with_gaps, n_prot_with_gaps / len(matches) * 100))
 
 
+def parse_unvalid_matches(matches, tsv):
+    """
+    Parse unvalid matches and write tsv summary file
+
+    :param matches: matches to parse
+    :type matches: list
+    :param tsv: output summary tsv file path
+    :type tsv: str
+    """
+    if os.path.exists(tsv):
+        os.remove(tsv)
+    with open(tsv, "w") as tsv_f:
+        tsv_f.write("\t".join(("Contig", "Protein", "prot_cov", "score", "e_value")) + "\n")
+        for match in matches:
+            cov = "%0.2f" % (abs(match["prot"]["end"] - match["prot"]["start"]) / match["prot"]["len"] * 100)
+            tsv_f.write("\t".join((match["ref"]["id"], match["prot"]["id"], cov, match["bitscore"], match["e_value"])) +
+                        "\n")
+
+
 if __name__ == "__main__":
     import argparse
 
@@ -352,4 +371,8 @@ if __name__ == "__main__":
 
     maps, bad_maps = get_valid_maps(args.diamond)
     print("load OK", flush=True)
+    print("Fix contigs...", flush=True)
     parse_valid_matches(maps, args.output_prefix, args.reference)
+    print("Summarize excluded matches...", flush=True)
+    parse_unvalid_matches(bad_maps, args.output_prefix + "_ignored_matches.tsv")
+    print("Done")