From 3049472c49a876370f7ce16d2421227843a98c69 Mon Sep 17 00:00:00 2001
From: ctbus <ctbushman@gmail.com>
Date: Thu, 19 Oct 2023 09:58:48 -0400
Subject: [PATCH] Filter on viral contig length

---
 sbx_virus_id.smk                    | 2 +-
 scripts/filter_virsorter_contigs.py | 6 ++++++
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/sbx_virus_id.smk b/sbx_virus_id.smk
index 8ca3ee7..ca4f8d4 100755
--- a/sbx_virus_id.smk
+++ b/sbx_virus_id.smk
@@ -228,7 +228,7 @@ rule virsorter:
         "envs/virsorter_env.yml"
     shell:
         """
-        virsorter run -w {params.out_dir} -i {input.contigs} -j {threads} all
+        virsorter run -w {params.out_dir} -i {input.contigs} --min-length 1000 -j {threads} all
         """
 
 
diff --git a/scripts/filter_virsorter_contigs.py b/scripts/filter_virsorter_contigs.py
index e69de29..89e28f7 100755
--- a/scripts/filter_virsorter_contigs.py
+++ b/scripts/filter_virsorter_contigs.py
@@ -0,0 +1,6 @@
+from sunbeamlib.parse import parse_fasta, write_fasta
+
+# This does nothing but I'm leaving it in case we want to add custom filtering here later
+with open(snakemake.input.contigs) as f_contigs, open(snakemake.output[0], "w") as f_out:
+    for header_str, seq_str in parse_fasta(f_contigs):
+        write_fasta((header_str, seq_str), f_out)
\ No newline at end of file