From 656d11fc77b2cb63d0db7667fb0cbde435c10fe6 Mon Sep 17 00:00:00 2001
From: Taisiia Sherstiukova <tarabulka@gmail.com>
Date: Wed, 2 Apr 2025 15:54:45 -0600
Subject: [PATCH] changed the sequence and proteinSequence functions to make
 all sequences upper case no matter the input format

---
 .DS_Store                               | Bin 6148 -> 6148 bytes
 excelutils/.DS_Store                    | Bin 6148 -> 6148 bytes
 excelutils/excel_sbol_utils/.DS_Store   | Bin 6148 -> 6148 bytes
 excelutils/excel_sbol_utils/library2.py |   8 ++++----
 excelutils/excel_sbol_utils/test.py     |   4 ++++
 5 files changed, 8 insertions(+), 4 deletions(-)
 create mode 100644 excelutils/excel_sbol_utils/test.py

diff --git a/.DS_Store b/.DS_Store
index f44001aebe3168b95c406d69715458c38c5d1a94..dd6e21a0facfee5ffd04bf2205a4a36c4b03811a 100644
GIT binary patch
delta 127
zcmZoMXffEJ#uU5iHv<C$3xgg*IzuKyNp8N2OHxjL5>Slev8?Rfy+@BZqROY>l`qIJ
r3{K9^Edc6aU{F0UxtZw(Q`N7@vCMUh37a1>tFcThVBO5l@s}R}7NI4c

delta 127
zcmZoMXffEJ#uO_f$H2hA!l1{H&XCDalAG`1l9ZF51Qg@Q=QjKBb^9?#RQVLV@&y@&
q!O8i#1wcIv3@QSXo0)DfmHeC>%Us78u=ydg8q352*3Il3fB6AB&Lax|

diff --git a/excelutils/.DS_Store b/excelutils/.DS_Store
index 23ddafc31379b030ea9b62af7816cb661003caac..e71403e653cb92f198f597b88f96d7b593628b3c 100644
GIT binary patch
delta 49
zcmZoMXffE3&B7F$G&z^WfP>MmgKPWIV~#+M+T_zLTx<}5$+uZd7!Pa~VErNl0G^@|
A7XSbN

delta 49
zcmZoMXffE3&B7F(I60TafaCtrt*6&-JLU-Fs7*f2!o>y=n0%YXgmJ@W0oE@<00yNM
ASO5S3

diff --git a/excelutils/excel_sbol_utils/.DS_Store b/excelutils/excel_sbol_utils/.DS_Store
index 0c1ddf2650ed3e2cf848f4a4217d11380326f19e..274bcf39e195eaa7bb06afb19ed620aa2a46e880 100644
GIT binary patch
delta 45
zcmZoMXffDe!pP*#Fqwx@YH|Q04_o_7o20!*Cr@CMnOq0tUd^uf^Wog)35+FT085Jy
AdjJ3c

delta 45
ucmZoMXffDe!pP)&cQOy7)Z_q09=5GjE^gNjPM*LhGr11V-8_M@L<|60oDcE<

diff --git a/excelutils/excel_sbol_utils/library2.py b/excelutils/excel_sbol_utils/library2.py
index c064d63..5754ac8 100644
--- a/excelutils/excel_sbol_utils/library2.py
+++ b/excelutils/excel_sbol_utils/library2.py
@@ -982,9 +982,9 @@ def sequence(rowobj):
 			elif re.match(r'^[a-zA-Z \s*]+$', val):
 				# if a sequence string
 
-				# removes spaces, enters, and makes all lower case
+				# removes spaces, enters, and makes all upper case
 				val = "".join(val.split())
-				val = val.replace(u"\ufeff", "").lower()
+				val = val.replace(u"\ufeff", "").upper()
 
 				uri = f'{url}/search/sequence={val}&'
 				if data["Domain"].strip() == "":
@@ -1042,10 +1042,10 @@ def proteinSequence(rowobj):
 					return
 				rowobj.obj.sequences = [val]
 
-			elif re.match(r'^[ACDEFGHIKLMNPQRSTVWY\s*]+$', val):
+			elif re.match(r'^[ACDEFGHIKLMNPQRSTVWYacdefghiklmnpqrstvwy\s*]+$', val):
 				# if a sequence string
 
-				# removes spaces, enters, and makes all lower case
+				# removes spaces, enters, and makes all upper case
 				val = "".join(val.split())
 				# removes *
 				val = val.replace('*', '')
diff --git a/excelutils/excel_sbol_utils/test.py b/excelutils/excel_sbol_utils/test.py
new file mode 100644
index 0000000..fdf168e
--- /dev/null
+++ b/excelutils/excel_sbol_utils/test.py
@@ -0,0 +1,4 @@
+user_input = "   Hello   "
+cleaned_input = user_input.strip()
+print(f"Original input: '{user_input}'")
+print(f"Cleaned input: '{cleaned_input}'")
\ No newline at end of file