From 6a7a2b177ab7ee17da75ad7cfcc097b6cd52e187 Mon Sep 17 00:00:00 2001
From: Vijini Mallawaarachchi <viji.mallawaarachchi@gmail.com>
Date: Thu, 13 Jun 2024 13:45:58 +0930
Subject: [PATCH] DEV: Add VOG count threshold fix #5

---
 reneo/__main__.py                                     | 10 +++++++++-
 reneo/config/config.yaml                              |  1 +
 reneo/workflow/rules/reneo.smk                        |  1 +
 reneo/workflow/scripts/reneo.py                       |  4 ++++
 reneo/workflow/scripts/reneo_utils/component_utils.py |  2 +-
 5 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/reneo/__main__.py b/reneo/__main__.py
index ba0d99b..1b56e1f 100644
--- a/reneo/__main__.py
+++ b/reneo/__main__.py
@@ -161,7 +161,7 @@ def run_options(func):
             "--evalue",
             default=1e-10,
             required=False,
-            help="maximum e-value for phrog annotations",
+            help="maximum e-value for vog annotations",
             type=float,
             show_default=True,
         ),
@@ -173,6 +173,14 @@ def run_options(func):
             type=float,
             show_default=True,
         ),
+        click.option(
+            "--nvogs",
+            default=10,
+            required=False,
+            help="minimum number of vogs to consider a component",
+            type=int,
+            show_default=True,
+        ),
         click.option(
             "--covtol",
             default=100,
diff --git a/reneo/config/config.yaml b/reneo/config/config.yaml
index 95eb257..f332cf6 100644
--- a/reneo/config/config.yaml
+++ b/reneo/config/config.yaml
@@ -25,6 +25,7 @@ reneo:
     mgfrac: 0.2
     evalue: 1E-10
     hmmscore: 50
+    nvogs: 10
     covtol: 100
     alpha: 1.2
     koverage_args:
diff --git a/reneo/workflow/rules/reneo.smk b/reneo/workflow/rules/reneo.smk
index deb5f26..fd6daeb 100644
--- a/reneo/workflow/rules/reneo.smk
+++ b/reneo/workflow/rules/reneo.smk
@@ -24,6 +24,7 @@ rule run_reneo:
         mgfrac = config['mgfrac'],
         evalue = config['evalue'],
         hmmscore = config['hmmscore'],
+        nvogs = config['nvogs'],
         covtol = config['covtol'],
         alpha = config['alpha'],
         output = RESDIR,
diff --git a/reneo/workflow/scripts/reneo.py b/reneo/workflow/scripts/reneo.py
index bf43162..57221ec 100644
--- a/reneo/workflow/scripts/reneo.py
+++ b/reneo/workflow/scripts/reneo.py
@@ -1360,6 +1360,9 @@ def main(**kwargs):
     kwargs["logger"].info(
         f"Minimum hmm score for vog annotations: {kwargs['hmmscore']}"
     )
+    kwargs["logger"].info(
+        f"Minimum number of vogs to consider a component: {kwargs['nvogs']}"
+    )
     kwargs["logger"].info(
         f"Coverage tolerance for extending subpaths: {kwargs['covtol']}"
     )
@@ -1622,6 +1625,7 @@ def main(**kwargs):
         mgfrac=float(snakemake.params.mgfrac),
         evalue=float(snakemake.params.evalue),
         hmmscore=float(snakemake.params.hmmscore),
+        nvogs = int(snakemake.params.nvogs),
         covtol=float(snakemake.params.covtol),
         alpha=float(snakemake.params.alpha),
         output=snakemake.params.output,
diff --git a/reneo/workflow/scripts/reneo_utils/component_utils.py b/reneo/workflow/scripts/reneo_utils/component_utils.py
index ba5669e..49f750e 100644
--- a/reneo/workflow/scripts/reneo_utils/component_utils.py
+++ b/reneo/workflow/scripts/reneo_utils/component_utils.py
@@ -41,7 +41,7 @@ def get_components(**kwargs):
 
             if (
                 vogs_present
-                # and unitig_names[unitig] in circular
+                and len(vogs_found) >= kwargs['nvogs']
                 and kwargs["edges_lengths"][kwargs["unitig_names"][unitig]] > kwargs["minlength"]
             ):
                 pruned_vs[i] = component