From 48986991ab5f8c29dae75ef540a21b1d79a3949b Mon Sep 17 00:00:00 2001
From: Vitaliy Mysak <vmysak@bccfe.ca>
Date: Wed, 31 May 2023 15:19:13 -0700
Subject: [PATCH] add --check-long-deletion option

---
 bin/proviral     | 12 ++++++++----
 intact/intact.py | 21 +++++++++++++++++++++
 2 files changed, 29 insertions(+), 4 deletions(-)

diff --git a/bin/proviral b/bin/proviral
index 077fefc..f1d5b61 100755
--- a/bin/proviral
+++ b/bin/proviral
@@ -46,6 +46,9 @@ def cli():
 @click.option(
     '--run-hypermut/--no-hypermut', default=False
 )
+@click.option(
+    '--check-long-deletion/--ignore-long-deletion', default=False
+)
 @click.option(
     '--include-small-orfs/--exclude-small-orfs', default=False)
 @click.option(
@@ -53,9 +56,9 @@ def cli():
     default=os.getcwd()
 )
 
-def intact(input_file, subtype, include_packaging_signal, 
-        include_rre, check_major_splice_donor_site, run_hypermut, 
-        include_small_orfs, working_folder):
+def intact(input_file, subtype, include_packaging_signal,
+           include_rre, check_major_splice_donor_site, run_hypermut,
+           check_long_deletion, include_small_orfs, working_folder):
     """
     Check consensus sequences for intactness.
     """
@@ -69,7 +72,8 @@ def intact(input_file, subtype, include_packaging_signal,
     try:
         intact_sequences, non_intact_sequences, orfs, errors = it.intact(
             folder, input_file, subtype, include_packaging_signal, include_rre,
-            check_major_splice_donor_site, run_hypermut, include_small_orfs
+            check_major_splice_donor_site, run_hypermut, check_long_deletion,
+            include_small_orfs
         )
         log.info('Intact sequences written to ' + intact_sequences)
         log.info('Non-intact sequences written to ' + non_intact_sequences)
diff --git a/intact/intact.py b/intact/intact.py
index 97cfd83..25d851b 100644
--- a/intact/intact.py
+++ b/intact/intact.py
@@ -107,7 +107,22 @@ def isHypermut(aln):
 #/end isHypermut
 
 
+def has_long_deletion(sequence, alignment):
+    """
+    Determines whether the sequence has a long deletion in it.
+    Keyword Args:
 
+        sequence -- the query sequence.
+        alignment -- multiple sequence alignment object containing the
+                     reference and query sequence.
+    """
+    # NOTE: This is the same check that HIVSeqInR uses.
+    if len(sequence.seq) < 8000:
+        return IntactnessError(sequence.id,
+                               LONGDELETION_ERROR,
+                               "Query sequence contains a long deletion.")
+    return None
+#/end has_long_deletion
 
 
 def has_mutated_major_splice_donor_site(alignment, 
@@ -554,6 +569,7 @@ def intact( working_dir,
             include_rre,
             check_major_splice_donor_site,
             run_hypermut,
+            check_long_deletion,
             include_small_orfs,
             hxb2_forward_orfs = const.DEFAULT_FORWARD_ORFs,
             hxb2_reverse_orfs = const.DEFAULT_REVERSE_ORFS,
@@ -667,6 +683,11 @@ def intact( working_dir,
                 if hypermutated is not None:
                     sequence_errors.append(hypermutated)
 
+            if check_long_deletion is not None:
+                long_deletion = has_long_deletion(sequence, alignment)
+                if long_deletion:
+                    sequence_errors.append(long_deletion)
+
             orfs[sequence.id] = hxb2_found_orfs
             if len(sequence_errors) == 0:
                 intact_sequences.append(sequence)