geney 1.4.27__py2.py3-none-any.whl → 1.4.29__py2.py3-none-any.whl

geney/utils/spliceai_pytorch_utils.py

@@ -0,0 +1,65 @@
+
+import absl.logging
+absl.logging.set_verbosity(absl.logging.ERROR)
+
+import sys
+import numpy as np
+
+
+import torch
+from spliceai_pytorch import SpliceAI
+model = SpliceAI.from_preconfigured('10k')
+
+
+if sys.platform == 'darwin':
+    device = torch.device("mps") if torch.backends.mps.is_available() else torch.device("cpu")
+
+if sys.platform == 'linux':
+    device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
+
+
+print(f"SpliceAI loaded to {device}.")
+model.to(device)
+
+def one_hot_encode(seq):
+
+    map = np.asarray([[0, 0, 0, 0],
+                      [1, 0, 0, 0],
+                      [0, 1, 0, 0],
+                      [0, 0, 1, 0],
+                      [0, 0, 0, 1]])
+
+    seq = seq.upper().replace('A', '\x01').replace('C', '\x02')
+    seq = seq.replace('G', '\x03').replace('T', '\x04').replace('N', '\x00')
+
+    return map[np.fromstring(seq, np.int8) % 5]
+
+
+def sai_predict_probs(seq: str, model) -> list:
+    '''
+    Predicts the donor and acceptor junction probability of each
+    NT in seq using SpliceAI.
+
+    Let m:=2*sai_mrg_context + L be the input seq length. It is assumed
+    that the input seq has the following structure:
+
+          seq = |<sai_mrg_context NTs><L NTs><sai_mrg_context NTs>|
+
+    The returned probability matrix is of size 2XL, where
+    the first row is the acceptor probability and the second row
+    is the donor probability. These probabilities corresponds to the
+    middel <L NTs> NTs of the input seq.
+    '''
+    x = one_hot_encode(seq)[None, :]
+    y = model(x)
+    y = y[0, :, 1:].T
+    return y[0, :], y[1, :]
+
+
+def run_spliceai_seq(seq, indices, threshold=0):
+    # seq = 'N' * 5000 + seq + 'N' * 5000
+    ref_seq_probs_temp = sai_predict_probs(seq, model)
+    ref_seq_acceptor_probs, ref_seq_donor_probs = ref_seq_probs_temp[0, :], ref_seq_probs_temp[1, :]
+    acceptor_indices = {a: b for a, b in list(zip(indices, ref_seq_acceptor_probs)) if b >= threshold}
+    donor_indices = {a: b for a, b in list(zip(indices, ref_seq_donor_probs)) if b >= threshold}
+    return donor_indices, acceptor_indices

geney/utils/splicing_utils.py

@@ -1,9 +1,10 @@
 # __all__ = ['run_splicing_engine', 'adjoin_splicing_outcomes', 'process_epistasis']
 
 import pandas as pd
-from typing import List, Tuple
+from typing import List, Tuple, Optional
 
-def run_splicing_engine(seq: str, engine: str = 'spliceai') -> Tuple[List[float], List[float]]:
+# def run_splicing_engine(seq: Optional[str] = None, engine: str = 'spliceai') -> Tuple[List[float], List[float]]:
+def run_splicing_engine(seq: Optional[str] = None, engine: str = 'spliceai') -> Tuple[List[float], List[float]]:
     """
     Run the specified splicing engine to predict splice site probabilities on a sequence.
 
@@ -17,11 +18,19 @@ def run_splicing_engine(seq: str, engine: str = 'spliceai') -> Tuple[List[float]
     Raises:
         ValueError: If the engine is not implemented.
     """
+
+    if seq is None:
+        from geney.util.utils import generate_random_sequence
+        seq = generate_random_sequence(15_000)
+
     match engine:
         case 'spliceai':
-            from geney.utils.spliceai_utils import sai_predict_probs, sai_models
-            # print(seq)
-            acceptor_probs, donor_probs = sai_predict_probs(seq, models=sai_models)
+            # from geney.utils.spliceai_utils import sai_predict_probs, sai_models
+            # acceptor_probs, donor_probs = sai_predict_probs(seq, models=sai_models)
+
+            from geney.utils.spliceai_pytorch_utils import sai_predict_probs, model
+            acceptor_probs, donor_probs = sai_predict_probs(seq, model=model)
+
         case 'pangolin':
             from geney.utils.pangolin_utils import pangolin_predict_probs, pang_models
             # print(seq)

geney/utils/utils.py

@@ -6,6 +6,7 @@ import json
 # from pathlib import Path
 from bisect import bisect_left
 import hashlib
+import random
 
 # def is_monotonic(A):
 #     x, y = [], []
@@ -60,6 +61,9 @@ def is_monotonic(A):
     return all(x <= y for x, y in zip(A, A[1:])) or all(x >= y for x, y in zip(A, A[1:]))
 
 
+def generate_random_sequence(length: int) -> str:
+    """Generates a random DNA sequence of given length (A, C, G, T)."""
+    return ''.join(random.choices('ACGT', k=length))
 
 def generate_random_nucleotide_sequences(num_sequences, min_len=3, max_len=10):
     """

{geney-1.4.27.dist-info → geney-1.4.29.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geney
-Version: 1.4.27
+Version: 1.4.29
 Summary: A Python package for gene expression modeling.
 Home-page: https://github.com/nicolaslynn/geney
 Author: Nicolas Lynn

{geney-1.4.27.dist-info → geney-1.4.29.dist-info}/RECORD

@@ -43,10 +43,11 @@ geney/utils/TranscriptLibrary.py,sha256=W1hv4Y8wRlmwTs3iFdn4_IqS-2suVDzZe4fwti2K
 geney/utils/__init__.py,sha256=-nJ-DMx1JzP-ZCe_QuQCeM0ZYIT_16jxoXDhUaO_4Oc,714
 geney/utils/mutation_utils.py,sha256=r-pHr56gEa5kh_DPX8MjFY3ZfYaOtyo4CUfJ5ZHlXPw,3243
 geney/utils/pangolin_utils.py,sha256=JQSPbWxdzqGFYfWQktkfLMaMSGR28eGQhNzO7MLMe5M,6162
+geney/utils/spliceai_pytorch_utils.py,sha256=PfMgaoG6ftWfqKZKc_JNqj5wqQRUR2B-4YF22-zNh1M,2079
 geney/utils/spliceai_utils.py,sha256=VtrIbjyQxk_3lw86eWjftRYyal9OzxArJ0GV5u_ymTg,2721
-geney/utils/splicing_utils.py,sha256=vPCGnCPR1ooEZEHR79yFHLmRQXEJHXEQjjxpBR-YWOs,20635
-geney/utils/utils.py,sha256=m51Vd0cEbrcIHo6_8BAuI9YSPcKRs22e5LfVd2Qj6Is,2181
-geney-1.4.27.dist-info/METADATA,sha256=m4-HDmqjW43J6QhNdel4Luy18FvL4FXAHw-zE1hPDcQ,990
-geney-1.4.27.dist-info/WHEEL,sha256=AHX6tWk3qWuce7vKLrj7lnulVHEdWoltgauo8bgCXgU,109
-geney-1.4.27.dist-info/top_level.txt,sha256=O-FuNUMb5fn9dhZ-dYCgF0aZtfi1EslMstnzhc5IIVo,6
-geney-1.4.27.dist-info/RECORD,,
+geney/utils/splicing_utils.py,sha256=wyO0CzNsfragDbAt9J016zxiA-MUuE2r9eyVAC0bZxk,21051
+geney/utils/utils.py,sha256=GXqlatNhix1akt3fburNzIwhiW9ZdCQSt2vmU80neyA,2370
+geney-1.4.29.dist-info/METADATA,sha256=DvrTqggg2aWWJ56k49iotD3zC0PO8VHWnvQwDxSPwho,990
+geney-1.4.29.dist-info/WHEEL,sha256=AHX6tWk3qWuce7vKLrj7lnulVHEdWoltgauo8bgCXgU,109
+geney-1.4.29.dist-info/top_level.txt,sha256=O-FuNUMb5fn9dhZ-dYCgF0aZtfi1EslMstnzhc5IIVo,6
+geney-1.4.29.dist-info/RECORD,,