Xianfish9 commited on
Commit
cf6e0f3
·
verified ·
1 Parent(s): ded76e3

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +17 -16
app.py CHANGED
@@ -59,55 +59,56 @@ def extract_features_from_seq(sequence_list):
59
  # --- 4. 核心预测函数 ---
60
  def predict(sequence_input):
61
  if model is None:
62
- # 如果模型加载失败,可以提前抛出错误
63
  raise gr.Error("模型未能加载或初始化失败,请检查后台日志。")
64
 
65
  if not sequence_input or not isinstance(sequence_input, str):
66
- # 对于无效输入,也直接抛出错误
67
  raise gr.Error("请输入有效的生物序列。")
68
 
69
  cleaned_sequence = sequence_input.strip().upper()
 
 
 
 
 
 
70
  sequence_list = [cleaned_sequence]
71
 
72
- # !!! 移除这里的 try...except !!!
73
- # 让任何可能发生的错误自然地被Gradio捕获
74
  x1_np, x2_np = extract_features_from_seq(sequence_list)
75
 
76
- # 将 NumPy 数组转换为 PyTorch 张量
77
  tensor_x1 = torch.tensor(x1_np).to(device)
78
  tensor_x2 = torch.tensor(x2_np).to(device)
79
 
80
- # 模型预测
81
  with torch.no_grad():
82
  outputs = model(tensor_x1, tensor_x2)
83
 
84
- # 计算概率
85
  probabilities = torch.sigmoid(outputs).squeeze().cpu().numpy()
86
 
87
- # 准备输出结果
88
  labels = ["类别 A (a)", "类别 C (c)", "类别 M (m)", "类别 S (s)"]
89
- # 确保即使只有一个序列,结果也能正确处理
90
- if probabilities.ndim == 0: # 如果只有一个输出
91
- probabilities = [probabilities]
92
-
93
  result = {label: float(prob) for label, prob in zip(labels, probabilities)}
94
 
95
  return result
96
 
97
  # --- 5. 创建并启动 Gradio 界面 ---
 
 
98
  demo = gr.Interface(
99
  fn=predict,
100
  inputs=gr.Textbox(
101
  lines=7,
102
  label="输入生物序列 (Input Sequence)",
103
- placeholder="请在这里粘贴你的序列..."
 
104
  ),
105
  outputs=gr.Label(num_top_classes=4, label="预测概率 (Prediction Probabilities)"),
106
  title="CAFN 模型部署:多标签序列分类器",
107
- description="输入一个生物序列,模型将预测它属于四个类别 (A, C, M, S) 中每一个的概率。",
 
 
108
  examples=[
109
- ["PLEPIPIVAAAAA"],
110
- ["GMWSGGGGISGSLIIVIRAELGVPSGMMILGYLN"],
 
111
  ]
112
  )
113
 
 
59
  # --- 4. 核心预测函数 ---
60
  def predict(sequence_input):
61
  if model is None:
 
62
  raise gr.Error("模型未能加载或初始化失败,请检查后台日志。")
63
 
64
  if not sequence_input or not isinstance(sequence_input, str):
 
65
  raise gr.Error("请输入有效的生物序列。")
66
 
67
  cleaned_sequence = sequence_input.strip().upper()
68
+
69
+ # --- 新增:在这里进行长度检查 ---
70
+ EXPECTED_LENGTH = 49 # 定义期望的序列长度
71
+ if len(cleaned_sequence) != EXPECTED_LENGTH:
72
+ raise gr.Error(f"输入序列长度错误!模型要求序列长度必须为 {EXPECTED_LENGTH} 个字符,但您输入的长度为 {len(cleaned_sequence)}。")
73
+
74
  sequence_list = [cleaned_sequence]
75
 
76
+ # 现在只有在长度正确时,才会执行下面的特征提取
 
77
  x1_np, x2_np = extract_features_from_seq(sequence_list)
78
 
 
79
  tensor_x1 = torch.tensor(x1_np).to(device)
80
  tensor_x2 = torch.tensor(x2_np).to(device)
81
 
 
82
  with torch.no_grad():
83
  outputs = model(tensor_x1, tensor_x2)
84
 
 
85
  probabilities = torch.sigmoid(outputs).squeeze().cpu().numpy()
86
 
 
87
  labels = ["类别 A (a)", "类别 C (c)", "类别 M (m)", "类别 S (s)"]
 
 
 
 
88
  result = {label: float(prob) for label, prob in zip(labels, probabilities)}
89
 
90
  return result
91
 
92
  # --- 5. 创建并启动 Gradio 界面 ---
93
+ valid_example_sequence = "CGKSFIWSSTLFKHKRIHTGEKPYKCEECGKAFNHSQILLHIRHKRMHT"[:49]# 简单用49个'A'作为示例,你可以替换成一个更有代表性的序列
94
+
95
  demo = gr.Interface(
96
  fn=predict,
97
  inputs=gr.Textbox(
98
  lines=7,
99
  label="输入生物序列 (Input Sequence)",
100
+ # 在占位符中提示长度要求
101
+ placeholder="请在这里粘贴长度为 49 的序列..."
102
  ),
103
  outputs=gr.Label(num_top_classes=4, label="预测概率 (Prediction Probabilities)"),
104
  title="CAFN 模型部署:多标签序列分类器",
105
+ # 在描述中明确强调长度要求
106
+ description="输入一个生物序列,模型将预测它属于四个类别 (A, C, M, S) 中每一个的概率。\n\n**重要提示:本模型要求输入的序列长度必须为 49 个字符。**",
107
+ # 提供一个或多个长度正确的示例
108
  examples=[
109
+ [valid_example_sequence],
110
+ # 如果有其他示例,也确保它们长度是49
111
+ # ["LFPYASLRRWHQNVQDLMVAIDNLQEFFSSLPKGLHLLLRLQFLPQSL"[:49]]
112
  ]
113
  )
114