inline individual sequence analysis

cfe-lab · Sep 27, 2023 · 8136533 · 8136533
1 parent baf4bf3
commit 8136533
Show file tree

Hide file tree

Showing 6 changed files with 711 additions and 713 deletions.
diff --git a/intact/intact.py b/intact/intact.py
@@ -722,8 +722,9 @@ def intact( working_dir,
         for sequence in SeqIO.parse(in_handle, "fasta"):
             subtype_choices[sequence.id] = sequence
 
-    def analyse_single_sequence(holistic, sequence, blast_rows):
+    def analyse_single_sequence(writer, sequence, blast_rows):
         sequence_errors = []
+        holistic = HolisticInfo()
 
         invalid_subsequences = find_invalid_subsequences(sequence)
         if invalid_subsequences:
@@ -871,19 +872,16 @@ def analyse_single_sequence(holistic, sequence, blast_rows):
         if not include_small_orfs:
             sequence_errors.extend(small_orf_errors)
 
-        return is_intact, hxb2_found_orfs, sequence_errors
+        orfs = [x.__dict__ for x in hxb2_found_orfs]
+        errors = [x.__dict__ for x in sequence_errors]
+        holistic = holistic.__dict__
+        writer.write(sequence, is_intact, orfs, errors, holistic)
 
     with OutputWriter(working_dir, "csv" if output_csv else "json") as writer:
 
         blast_it = blast_iterate_inf(subtype, input_file, working_dir) if check_internal_inversion or check_nonhiv or check_scramble or 1 < len(subtype_choices) else iterate_empty_lists()
         for (sequence, blast_rows) in with_blast_rows(blast_it, iterate_sequences(input_file)):
-            holistic = HolisticInfo()
-            is_intact, hxb2_found_orfs, sequence_errors = analyse_single_sequence(holistic, sequence, blast_rows)
-            orfs = [x.__dict__ for x in hxb2_found_orfs]
-            errors = [x.__dict__ for x in sequence_errors]
-            holistic = holistic.__dict__
-            writer.write(sequence, is_intact, orfs, errors, holistic=holistic)
-
+            analyse_single_sequence(writer, sequence, blast_rows)
 
 #/end def intact
 #/end intact.py
diff --git a/tests/expected-results-edgy/errors.json b/tests/expected-results-edgy/errors.json
@@ -1,83 +1,83 @@
 {
     "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455": [],
-    "small-sequence": [
+    "small-sequence[REVERSE_COMPLEMENT]": [
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "ORF gag at 1-1498 can have maximum deletions 30, got 1494"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "ORF pol at 1290-4302 can have maximum deletions 30, got 2988"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "ORF env at 5430-8007 can have maximum deletions 100, got 2574"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF vif at 4246-4822 can have maximum deletions 30, got 573"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF vpr at 4764-5052 can have maximum deletions 30, got 285"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF tat_exon1 at 5032-5248 can have maximum deletions 30, got 216"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF rev_exon1 at 5171-5249 can have maximum deletions 30, got 78"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF vpu at 5267-5513 can have maximum deletions 30, got 243"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF tat_exon2 at 7567-7663 can have maximum deletions 30, got 90"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF rev_exon2 at 7568-7865 can have maximum deletions 30, got 294"
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF nef at 8008-8683 can have maximum deletions 30, got 672"
         },
         {
-            "sequence_name": "small-sequence",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "PackagingSignalDeletion",
             "message": "Query Sequence exceeds maximum deletion tolerance in PSI. Contains 21 deletions with max tolerance of 10 deletions."
         },
         {
-            "sequence_name": "small-sequence",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "RevResponseElementDeletion",
             "message": "Query Sequence exceeds maximum deletion tolerance in RRE. Contains 265 deletions with max tolerance of 20 deletions."
         },
         {
-            "sequence_name": "small-sequence",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "MajorSpliceDonorSiteMutated",
             "message": "Query sequence has a missing splice donor site, -."
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "LongDeletion",
             "message": "Query sequence contains a long deletion."
         },
         {
-            "sequence_name": "small-sequence [REVERSED]",
+            "sequence_name": "small-sequence[REVERSE_COMPLEMENT]",
             "error": "NonHIV",
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }
@@ -318,41 +318,41 @@
             "message": "Query sequence contains a long deletion."
         }
     ],
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED": [
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED[REVERSE_COMPLEMENT]": [
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF nef at 8008-8683 can have maximum deletions 30, got 54"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED[REVERSE_COMPLEMENT]",
             "error": "MajorSpliceDonorSiteMutated",
             "message": "Query sequence has a mutated splice donor site, G."
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED[REVERSE_COMPLEMENT]",
             "error": "NonHIV",
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }
     ],
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS": [
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS[REVERSE_COMPLEMENT]": [
         {
             "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS",
             "error": "UnknownNucleotide",
             "message": "Sequence contains invalid nucleotides: XXXXXXX (start: 45, end: 51)"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF nef at 8008-8683 can have maximum deletions 30, got 54"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS[REVERSE_COMPLEMENT]",
             "error": "MajorSpliceDonorSiteMutated",
             "message": "Query sequence has a mutated splice donor site, G."
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS[REVERSE_COMPLEMENT]",
             "error": "NonHIV",
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }
@@ -444,24 +444,24 @@
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }
     ],
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS": [
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS[REVERSE_COMPLEMENT]": [
         {
             "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS",
             "error": "UnknownNucleotide",
             "message": "Sequence contains invalid nucleotides: >1>#>@> (start: 45, end: 51)"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF nef at 8008-8683 can have maximum deletions 30, got 54"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS[REVERSE_COMPLEMENT]",
             "error": "MajorSpliceDonorSiteMutated",
             "message": "Query sequence has a mutated splice donor site, G."
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS[REVERSE_COMPLEMENT]",
             "error": "NonHIV",
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }
@@ -553,24 +553,24 @@
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }
     ],
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES": [
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES[REVERSE_COMPLEMENT]": [
         {
             "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES",
             "error": "UnknownNucleotide",
             "message": "Sequence contains invalid nucleotides: ------- (start: 45, end: 51)"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES[REVERSE_COMPLEMENT]",
             "error": "DeletionInOrf",
             "message": "Smaller ORF nef at 8008-8683 can have maximum deletions 30, got 54"
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES[REVERSE_COMPLEMENT]",
             "error": "MajorSpliceDonorSiteMutated",
             "message": "Query sequence has a mutated splice donor site, G."
         },
         {
-            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES [REVERSED]",
+            "sequence_name": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES[REVERSE_COMPLEMENT]",
             "error": "NonHIV",
             "message": "Sequence contains unrecognized parts. It is probably a Human/HIV Chimera sequence."
         }

diff --git a/tests/expected-results-edgy/holistic.json b/tests/expected-results-edgy/holistic.json
@@ -11,7 +11,7 @@
         "orfs_end": 8794,
         "blast_n_conseqs": 3
     },
-    "small-sequence": {
+    "small-sequence[REVERSE_COMPLEMENT]": {
         "qlen": 19,
         "hypermutation_probablility": 0.0,
         "inferred_subtype": "Ref.01_AE.AF.07.569M.GQ477441",
@@ -59,7 +59,7 @@
         "orfs_end": 8007,
         "blast_n_conseqs": 0
     },
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED": {
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.REVERSED[REVERSE_COMPLEMENT]": {
         "qlen": 9718,
         "hypermutation_probablility": 0.13527282947774355,
         "inferred_subtype": "Ref.01_AE.AF.07.569M.GQ477441",
@@ -71,7 +71,7 @@
         "orfs_end": 8007,
         "blast_n_conseqs": 0
     },
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS": {
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-XS[REVERSE_COMPLEMENT]": {
         "qlen": 9711,
         "hypermutation_probablility": 0.13527282947774355,
         "inferred_subtype": "Ref.01_AE.AF.07.569M.GQ477441",
@@ -95,7 +95,7 @@
         "orfs_end": 8007,
         "blast_n_conseqs": 0
     },
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS": {
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-BAD-SYMBOLS[REVERSE_COMPLEMENT]": {
         "qlen": 9711,
         "hypermutation_probablility": 0.13527282947774355,
         "inferred_subtype": "Ref.01_AE.AF.07.569M.GQ477441",
@@ -119,7 +119,7 @@
         "orfs_end": 8007,
         "blast_n_conseqs": 0
     },
-    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES": {
+    "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455.BUT.WITH-DASHES[REVERSE_COMPLEMENT]": {
         "qlen": 9711,
         "hypermutation_probablility": 0.13527282947774355,
         "inferred_subtype": "Ref.01_AE.AF.07.569M.GQ477441",