From f3d664a4df4c62f7717769ac38851ad1629fd385 Mon Sep 17 00:00:00 2001
From: Gerry Tonkin-Hill <g.tonkinhill@gmail.com>
Date: Tue, 24 Sep 2024 02:54:13 +0200
Subject: [PATCH] exposed cdhit length requirement in gene family clustering
 steps to address #299

---
 panaroo/__main__.py      | 7 +++++++
 panaroo/clean_network.py | 3 +++
 2 files changed, 10 insertions(+)

diff --git a/panaroo/__main__.py b/panaroo/__main__.py
index 72ce87c..0e51864 100755
--- a/panaroo/__main__.py
+++ b/panaroo/__main__.py
@@ -114,6 +114,10 @@ def get_options(args):
                           dest="len_dif_percent",
                           help="length difference cutoff (default=0.98)",
                           type=float)
+    matching.add_argument("--family_len_dif_percent",
+                          dest="family_len_dif_percent",
+                          help="length difference cutoff at the gene family level (default=0.0)",
+                          type=float)
     matching.add_argument("--merge_paralogs",
                           dest="merge_paralogs",
                           help="don't split paralogs",
@@ -386,6 +390,7 @@ def main():
                           outdir=temp_dir,
                           dna_error_threshold=0.98,
                           correct_mistranslations=True,
+                          family_len_dif_percent=args.family_len_dif_percent,
                           length_outlier_support_proportion=args.
                           length_outlier_support_proportion,
                           n_cpu=args.n_cpu,
@@ -401,6 +406,7 @@ def main():
         outdir=temp_dir,
         family_threshold=args.family_threshold,
         correct_mistranslations=False,
+        family_len_dif_percent=args.family_len_dif_percent,
         length_outlier_support_proportion=args.
         length_outlier_support_proportion,
         n_cpu=args.n_cpu,
@@ -454,6 +460,7 @@ def main():
                             outdir=temp_dir,
                             family_threshold=args.family_threshold,
                             correct_mistranslations=False,
+                            family_len_dif_percent=args.family_len_dif_percent,
                             length_outlier_support_proportion=args.
                             length_outlier_support_proportion,
                             n_cpu=args.n_cpu,
diff --git a/panaroo/clean_network.py b/panaroo/clean_network.py
index a9312e2..0a52d07 100755
--- a/panaroo/clean_network.py
+++ b/panaroo/clean_network.py
@@ -92,6 +92,7 @@ def collapse_families(G,
                       outdir,
                       family_threshold=0.7,
                       dna_error_threshold=0.99,
+                      family_len_dif_percent=0,
                       correct_mistranslations=False,
                       length_outlier_support_proportion=0.01,
                       n_cpu=1,
@@ -113,6 +114,7 @@ def collapse_families(G,
         cdhit_clusters = iterative_cdhit(G,
                                          outdir,
                                          thresholds=threshold,
+                                         s=family_len_dif_percent,
                                          n_cpu=n_cpu,
                                          quiet=True,
                                          dna=True,
@@ -124,6 +126,7 @@ def collapse_families(G,
         cdhit_clusters = iterative_cdhit(G,
                                          outdir,
                                          thresholds=threshold,
+                                         s=family_len_dif_percent,
                                          n_cpu=n_cpu,
                                          quiet=True,
                                          dna=False)