Spaces:
Sleeping
Sleeping
Kevin Louis
commited on
Commit
•
7c244fe
1
Parent(s):
ed15e3a
Add application file
Browse files
app.py
ADDED
@@ -0,0 +1,175 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
|
3 |
+
import pandas as pd
|
4 |
+
from datasets import Dataset
|
5 |
+
from sentence_transformers import SentenceTransformer
|
6 |
+
from parameter_extractor import ParameterExtractor
|
7 |
+
from DNAseq import DNAseq
|
8 |
+
from helper import list_at_index_0, list_at_index_1, logger
|
9 |
+
|
10 |
+
|
11 |
+
def chat_to_sequence(sequence, user_query):
|
12 |
+
if sequence is None:
|
13 |
+
gr.Warning("Sequence Is Empty. Please Input A Sequence")
|
14 |
+
if user_query is None:
|
15 |
+
gr.Warning("Query Is Empty. Please Input A Query")
|
16 |
+
# Log information to a CSV file
|
17 |
+
log_filename = "CTS_user_log.csv"
|
18 |
+
|
19 |
+
# Sequence to be analysed/queried
|
20 |
+
input_sequence = sequence
|
21 |
+
|
22 |
+
# Set ParameterExtractor class expected variable
|
23 |
+
dna = input_sequence
|
24 |
+
|
25 |
+
# Model
|
26 |
+
model_name = "all-mpnet-base-v2"
|
27 |
+
|
28 |
+
# Load model
|
29 |
+
model = SentenceTransformer(model_name)
|
30 |
+
|
31 |
+
# User input
|
32 |
+
user_query = user_query
|
33 |
+
|
34 |
+
# Set ParameterExtractor class expected variable
|
35 |
+
query = user_query
|
36 |
+
|
37 |
+
# Bot Response
|
38 |
+
response = ""
|
39 |
+
|
40 |
+
# Query Code Description Message
|
41 |
+
code_descript_message = ''
|
42 |
+
|
43 |
+
# kNN semantic similarity threshold / used to determine if query can execute code
|
44 |
+
# kNN semantic similarity values less than the lower threshold should return a code eval response
|
45 |
+
# kNN semantic similarity values more than the lower threshold shouldn't return a code eval response
|
46 |
+
proximal_lower_threshold = 1.1
|
47 |
+
proximal_upper_threshold = 1.4
|
48 |
+
|
49 |
+
threshold_exceeded_message = "Your Query Wasn't Understood. Can You Rephrase The Query"
|
50 |
+
threshold_approximate_message = "Your Query Wasn't Understood Clearly. Try Using The Following Query Formats"
|
51 |
+
|
52 |
+
# Load the function mapping CSV file into a pandas DataFrame
|
53 |
+
code_function_mapping = pd.read_csv("code_function_mapping.csv")
|
54 |
+
|
55 |
+
# Load reference query database from JSON file back into a DataFrame
|
56 |
+
ref_query_df = pd.read_json('reference_query_db.json', orient='records')
|
57 |
+
|
58 |
+
# Create Dataset object using the pandas data frame
|
59 |
+
ref_query_ds = Dataset.from_pandas(ref_query_df)
|
60 |
+
|
61 |
+
# Load FAISS index
|
62 |
+
ref_query_ds.load_faiss_index('all-mpnet-base-v2_embeddings', 'ref_query_db_index')
|
63 |
+
|
64 |
+
# Create embeddings for user query
|
65 |
+
query_embedding = model.encode(user_query)
|
66 |
+
|
67 |
+
# Semantic similarity search user query against sample queries
|
68 |
+
index_result = ref_query_ds.get_nearest_examples("all-mpnet-base-v2_embeddings", query_embedding, k=3)
|
69 |
+
print(index_result)
|
70 |
+
|
71 |
+
# Retrieve results from dataset object
|
72 |
+
scores, examples = index_result
|
73 |
+
|
74 |
+
# Create a DataFrame from the examples dictionary
|
75 |
+
result_df = pd.DataFrame(examples)
|
76 |
+
|
77 |
+
# Add the scores as a new column to the DataFrame
|
78 |
+
result_df['score'] = scores
|
79 |
+
|
80 |
+
# Sort the DataFrame by the 'Score' column in ascending order
|
81 |
+
# FIASS uses kNN as the similarity algorithm / value of 0 indicates an exact match
|
82 |
+
sorted_df = result_df.sort_values(by='score', ascending=True)
|
83 |
+
|
84 |
+
# Get the query with the lowest kNN score (first row after sorting)
|
85 |
+
ref_question = sorted_df.iloc[0]['question']
|
86 |
+
|
87 |
+
# Get the code for the query with the lowest kNN score (first row after sorting)
|
88 |
+
query_code = sorted_df.iloc[0]['code']
|
89 |
+
|
90 |
+
# Get the score for the query with the lowest kNN score (first row after sorting)
|
91 |
+
query_score = sorted_df.iloc[0]['score']
|
92 |
+
|
93 |
+
# Description of query code to be executed
|
94 |
+
query_code_description = code_function_mapping[code_function_mapping['code'] == query_code]['description'].values[0]
|
95 |
+
|
96 |
+
# Print the query with the highest score
|
97 |
+
print(ref_question, query_code, query_score)
|
98 |
+
similarity_metric = "k nearest neighbours"
|
99 |
+
|
100 |
+
ref_question_2 = sorted_df.iloc[1]['question']
|
101 |
+
ref_question_3 = sorted_df.iloc[1]['question']
|
102 |
+
query_score_2 = sorted_df.iloc[1]['score']
|
103 |
+
query_score_3 = sorted_df.iloc[1]['score']
|
104 |
+
|
105 |
+
log_data = [
|
106 |
+
user_query,
|
107 |
+
ref_question,
|
108 |
+
query_score,
|
109 |
+
query_code,
|
110 |
+
ref_question_2,
|
111 |
+
query_score_2,
|
112 |
+
ref_question_3,
|
113 |
+
query_score_3,
|
114 |
+
similarity_metric,
|
115 |
+
model_name,
|
116 |
+
proximal_lower_threshold,
|
117 |
+
proximal_upper_threshold,
|
118 |
+
]
|
119 |
+
# Check the query score against threshold values
|
120 |
+
if query_score >= proximal_upper_threshold:
|
121 |
+
response = threshold_exceeded_message
|
122 |
+
logger(log_filename, log_data, response)
|
123 |
+
print(threshold_exceeded_message)
|
124 |
+
|
125 |
+
elif proximal_lower_threshold < query_score < proximal_upper_threshold:
|
126 |
+
response = threshold_approximate_message + "/n" + ref_question
|
127 |
+
logger(log_filename, log_data, response)
|
128 |
+
print(threshold_approximate_message, ref_question)
|
129 |
+
else:
|
130 |
+
print("Execute query")
|
131 |
+
# Define the question
|
132 |
+
code = query_code
|
133 |
+
|
134 |
+
# Filter the DataFrame to find the code that matches the question
|
135 |
+
matching_row = code_function_mapping[code_function_mapping["code"] == code]
|
136 |
+
|
137 |
+
# Check if there is a match
|
138 |
+
if not matching_row.empty:
|
139 |
+
function = matching_row.iloc[0]["function"]
|
140 |
+
response = str(eval(function))
|
141 |
+
code_descript_message = query_code_description.title()
|
142 |
+
logger(log_filename, log_data, response)
|
143 |
+
else:
|
144 |
+
response = "Error processing query"
|
145 |
+
query_code = "No Match Error"
|
146 |
+
logger(log_filename, log_data, response)
|
147 |
+
print("No matching code found for the function:", code)
|
148 |
+
|
149 |
+
return response, code_descript_message
|
150 |
+
return response, code_descript_message
|
151 |
+
|
152 |
+
|
153 |
+
ChatToSequence = gr.Interface(
|
154 |
+
fn=chat_to_sequence,
|
155 |
+
inputs=[gr.Textbox(label="Sequence", placeholder="Input DNA Sequence..."),
|
156 |
+
gr.Textbox(label="Query", placeholder="Input Query...")],
|
157 |
+
outputs=[gr.Textbox(label="Response"), gr.Textbox(label="Action Executed")],
|
158 |
+
title="Chat-To-Sequence",
|
159 |
+
description="This Demo App Allows You To Explore Your DNA Sequence Using Natural Language",
|
160 |
+
theme=gr.themes.Soft(),
|
161 |
+
examples=[
|
162 |
+
["ggcattgaggagaccattgacaccgtcattagcaatgcactacaactgtcacaacctaaa",
|
163 |
+
"What is the length of the sequence"],
|
164 |
+
["ggcattgaggagaccattgacaccgtcattagcaatgcactacaactgtcacaacctaaa",
|
165 |
+
"How many guanines bases are there in the sequence"],
|
166 |
+
["ggcattgaggagaccattgacaccgtcattagcaatgcactacaactgtcacaacctaaa",
|
167 |
+
"What is the base at position 10"],
|
168 |
+
["ggcattgaggagaccattgacaccgtcattagcaatgcactacaactgtcacaacctaaa",
|
169 |
+
"What are the bases from position 2 to 10"],
|
170 |
+
["ggcattgaggagaccattgacaccgtcattagcaatgcactacaactgtcacaacctaaa",
|
171 |
+
"How many bases are there from position 2 to 10"],
|
172 |
+
],
|
173 |
+
).queue()
|
174 |
+
|
175 |
+
ChatToSequence.launch(share=True)
|