From 568a6acc5959b843df96bdd659049691e65e2c6d Mon Sep 17 00:00:00 2001
From: desafinadude <meerkdir@gmail.com>
Date: Sat, 6 Dec 2025 19:33:06 +0200
Subject: [PATCH 1/3] handling fasta files with biopython

---
 helpers.py       | 50 +++++++++++++++++++++++++-----------------------
 requirements.txt |  1 +
 worker.py        |  2 ++
 3 files changed, 29 insertions(+), 24 deletions(-)

diff --git a/helpers.py b/helpers.py
index c16fe27..0d0ee6e 100644
--- a/helpers.py
+++ b/helpers.py
@@ -23,6 +23,8 @@
 )
 import base64
 from minio import Minio
+from Bio import SeqIO
+from io import StringIO
 
 from database import get_db_cursor
 import os
@@ -844,34 +846,34 @@ async def check_for_sequence_data(isolate, split_on_fasta_headers=True):
             socket.setdefaulttimeout(original_timeout)  # Restore timeout even on error
             return False, f"Error loading FASTA file from MinIO: {str(e)}"
         
-        # 5. Parse the FASTA file to check if header is in the file
-        fasta_lines = fasta_content.splitlines()
-        sequence_lines = []
-        recording = False
-        header_found = False
-        
-        for line in fasta_lines:
-            if line.startswith('>'):
-                # Check if this header matches what we're looking for
-                if line.startswith(f'>{fasta_header} ') or line == f'>{fasta_header}' or line.startswith(f'>{fasta_header}\t'):
-                    recording = True
-                    header_found = True
-                    sequence_lines.append(line)
-                else:
-                    # If we were recording and hit a different header, stop
-                    if recording:
-                        break
-            else:
-                if recording:
-                    sequence_lines.append(line)
-
-        print(sequence_lines)
+        # 5. Parse the FASTA file using BioPython to check if header is in the file
+        fasta_handle = StringIO(fasta_content)
+        
+        sequence_found = None
+        try:
+            for record in SeqIO.parse(fasta_handle, "fasta"):
+                # BioPython's record.id is the header without '>'
+                # record.description contains the full header line
+                # Check various header match patterns
+                if (record.id == fasta_header or 
+                    record.description == fasta_header or
+                    record.description.startswith(fasta_header + ' ') or
+                    record.description.startswith(fasta_header + '\t')):
+                    sequence_found = record
+                    break
+        except ValueError as e:
+            return False, f"Invalid FASTA format: {str(e)}"
         
         # 6. Return error if header not found
-        if not header_found:
+        if not sequence_found:
             return False, f"Header '{fasta_header}' not found in {fasta_file} for isolate '{isolate_sample_id}'"
         
-        sequence_data = '\n'.join(sequence_lines)
+        # Validate sequence isn't empty
+        if len(sequence_found.seq) == 0:
+            return False, f"Empty sequence found for header: {fasta_header}"
+        
+        # Reconstruct FASTA format for storage
+        sequence_data = f">{sequence_found.description}\n{str(sequence_found.seq)}"
         
         if not sequence_data.strip():
             return False, f"No sequence data found for isolate '{isolate_sample_id}'"
diff --git a/requirements.txt b/requirements.txt
index eb5dc1b..5d3df7c 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -9,3 +9,4 @@ pytest==7.4.3
 sendgrid==6.12.5
 minio==7.2.8
 pip_system_certs==5.3
+biopython==1.84
diff --git a/worker.py b/worker.py
index f6bb596..fbb7e47 100644
--- a/worker.py
+++ b/worker.py
@@ -3,6 +3,8 @@
 import sys
 import json
 import asyncio
+from io import StringIO
+from Bio import SeqIO
 from jobs import get_next_job, mark_job_done, mark_job_failed
 from database import get_db_cursor
 from helpers import check_for_sequence_data, send_to_elastic2

From ea2b437ee03a0df302b48bd2783d1c41bd5a1419 Mon Sep 17 00:00:00 2001
From: Di M <meerkdir@gmail.com>
Date: Mon, 8 Dec 2025 12:09:12 +0200
Subject: [PATCH 2/3] Update worker.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---
 worker.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/worker.py b/worker.py
index fbb7e47..ad36d77 100644
--- a/worker.py
+++ b/worker.py
@@ -4,7 +4,6 @@
 import json
 import asyncio
 from io import StringIO
-from Bio import SeqIO
 from jobs import get_next_job, mark_job_done, mark_job_failed
 from database import get_db_cursor
 from helpers import check_for_sequence_data, send_to_elastic2

From e7419821388e44d005f6b4c179abe9bc5c42084e Mon Sep 17 00:00:00 2001
From: Di M <meerkdir@gmail.com>
Date: Mon, 8 Dec 2025 12:09:27 +0200
Subject: [PATCH 3/3] Update helpers.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---
 helpers.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/helpers.py b/helpers.py
index ec266f0..2b09ff5 100644
--- a/helpers.py
+++ b/helpers.py
@@ -27,7 +27,6 @@
 import base64
 from minio import Minio
 from Bio import SeqIO
-from io import StringIO
 
 from database import get_db_cursor
 import os