File size: 6,089 Bytes
5263bd3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cdd8a58
 
 
 
 
 
 
 
 
 
 
 
 
 
5263bd3
9d48283
 
 
870813f
9d48283
 
 
 
 
 
870813f
 
 
9d48283
 
 
 
 
 
 
 
5263bd3
870813f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
63d967d
 
870813f
 
63d967d
723da6d
 
 
 
 
 
 
 
cdd8a58
 
 
 
 
870813f
723da6d
 
9d48283
 
 
 
 
 
 
723da6d
9d48283
 
723da6d
 
9d48283
723da6d
870813f
 
723da6d
 
 
 
 
 
cdd8a58
723da6d
041d3b3
 
 
cdd8a58
723da6d
cdd8a58
723da6d
 
cdd8a58
 
 
 
041d3b3
 
cdd8a58
 
 
 
 
 
 
 
041d3b3
cdd8a58
 
 
 
723da6d
 
 
cdd8a58
723da6d
5263bd3
 
 
cdd8a58
 
 
 
 
233c5f7
cdd8a58
723da6d
 
 
 
 
5263bd3
870813f
5263bd3
870813f
723da6d
870813f
 
5263bd3
 
723da6d
 
cdd8a58
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
import gradio as gr
import torch
import joblib
import numpy as np
from itertools import product
import torch.nn as nn

class VirusClassifier(nn.Module):
    def __init__(self, input_shape: int):
        super(VirusClassifier, self).__init__()
        self.network = nn.Sequential(
            nn.Linear(input_shape, 64),
            nn.GELU(),
            nn.BatchNorm1d(64),
            nn.Dropout(0.3),
            nn.Linear(64, 32),
            nn.GELU(),
            nn.BatchNorm1d(32),
            nn.Dropout(0.3),
            nn.Linear(32, 32),
            nn.GELU(),
            nn.Linear(32, 2)
        )

    def forward(self, x):
        return self.network(x)
    
    def get_feature_importance(self, x):
        """Calculate feature importance using gradient-based method"""
        x.requires_grad_(True)
        output = self.network(x)
        importance = torch.zeros_like(x)
        
        for i in range(output.shape[1]):
            if x.grad is not None:
                x.grad.zero_()
            output[..., i].sum().backward(retain_graph=True)
            importance += torch.abs(x.grad)
            
        return importance

def sequence_to_kmer_vector(sequence: str, k: int = 4) -> np.ndarray:
    """Convert sequence to k-mer frequency vector"""
    # Generate all possible k-mers
    kmers = [''.join(p) for p in product("ACGT", repeat=k)]
    kmer_dict = {km: i for i, km in enumerate(kmers)}

    # Initialize vector
    vec = np.zeros(len(kmers), dtype=np.float32)
    
    # Count k-mers
    for i in range(len(sequence) - k + 1):
        kmer = sequence[i:i+k]
        if kmer in kmer_dict:
            vec[kmer_dict[kmer]] += 1

    # Convert to frequencies
    total_kmers = len(sequence) - k + 1
    if total_kmers > 0:
        vec = vec / total_kmers

    return vec

def parse_fasta(text):
    sequences = []
    current_header = None
    current_sequence = []
    
    for line in text.split('\n'):
        line = line.strip()
        if not line:
            continue
        if line.startswith('>'):
            if current_header:
                sequences.append((current_header, ''.join(current_sequence)))
            current_header = line[1:]
            current_sequence = []
        else:
            current_sequence.append(line.upper())
    if current_header:
        sequences.append((current_header, ''.join(current_sequence)))
    return sequences

def predict(file_obj):
    if file_obj is None:
        return "Please upload a FASTA file"
    
    # Read the file content
    try:
        if isinstance(file_obj, str):
            text = file_obj
        else:
            text = file_obj.decode('utf-8')
    except Exception as e:
        return f"Error reading file: {str(e)}"

    # Generate k-mer dictionary
    k = 4  # k-mer size
    kmers = [''.join(p) for p in product("ACGT", repeat=k)]
    kmer_dict = {km: i for i, km in enumerate(kmers)}
    
    # Load model and scaler
    try:
        device = 'cuda' if torch.cuda.is_available() else 'cpu'
        model = VirusClassifier(256).to(device)  # k=4 -> 4^4 = 256 features
        
        # Load model with explicit map_location
        state_dict = torch.load('model.pt', map_location=device)
        model.load_state_dict(state_dict)
        
        # Load scaler
        scaler = joblib.load('scaler.pkl')
        
        # Set model to evaluation mode
        model.eval()
    except Exception as e:
        return f"Error loading model: {str(e)}\nFull traceback: {str(e.__traceback__)}"

    # Get predictions
    results = []
    try:
        sequences = parse_fasta(text)
        for header, seq in sequences:
            # Get k-mer vector
            kmer_vector = sequence_to_kmer_vector(seq)
            kmer_vector = scaler.transform(kmer_vector.reshape(1, -1))
            X_tensor = torch.FloatTensor(kmer_vector).to(device)
            
            # Get raw frequency vector before scaling
            raw_freq_vector = sequence_to_kmer_vector(seq)
            
            # Get predictions and feature importance
            with torch.no_grad():
                output = model(X_tensor)
                probs = torch.softmax(output, dim=1)
            
            # Calculate feature importance
            importance = model.get_feature_importance(X_tensor)
            kmer_importance = importance[0].cpu().numpy()
            
            # Normalize importance scores to match original scale
            kmer_importance = kmer_importance / np.max(np.abs(kmer_importance)) * 0.002
            
            # Get top 10 k-mers
            top_k = 10
            top_indices = np.argsort(np.abs(kmer_importance))[-top_k:][::-1]
            important_kmers = [
                {
                    'kmer': list(kmer_dict.keys())[list(kmer_dict.values()).index(i)],
                    'importance': float(kmer_importance[i]),
                    'frequency': float(raw_freq_vector[i])
                }
                for i in top_indices
            ]
            
            # Format results
            pred_class = 1 if probs[0][1] > probs[0][0] else 0
            pred_label = 'human' if pred_class == 1 else 'non-human'
            
            result = f"""Sequence: {header}
Prediction: {pred_label}
Confidence: {float(max(probs[0])):0.4f}
Human probability: {float(probs[0][1]):0.4f}
Non-human probability: {float(probs[0][0]):0.4f}

Most influential k-mers:"""
            
            for kmer in important_kmers:
                result += f"\n  {kmer['kmer']}: importance={kmer['importance']:.4f}, raw_freq={kmer['raw_freq']:.4f} ({kmer['raw_freq']*100:.2f}%), scaled_freq={kmer['scaled_freq']:.4f}"
            
            results.append(result)
    except Exception as e:
        return f"Error processing sequences: {str(e)}"

    return "\n\n".join(results)

# Create the interface
iface = gr.Interface(
    fn=predict,
    inputs=gr.File(label="Upload FASTA file", type="binary"),
    outputs=gr.Textbox(label="Results"),
    title="Virus Host Classifier"
)

# Launch the interface
if __name__ == "__main__":
    iface.launch()