seqchromloader 0.7.3__tar.gz → 0.7.4__tar.gz

{seqchromloader-0.7.3 → seqchromloader-0.7.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: seqchromloader
-Version: 0.7.3
+Version: 0.7.4
 Summary: Sequence and chromatin dataloader for deep learning
 Home-page: https://github.com/yztxwd/seqchromloader
 Author-email: yztxwd@gmail.com

{seqchromloader-0.7.3 → seqchromloader-0.7.4}/seqchromloader/loader.py

@@ -110,7 +110,8 @@ class _SeqChromDatasetByDataFrame(IterableDataset):
                  bigwig_filelist:list, 
                  target_bam=None, 
                  transforms:dict=None, 
-                 return_region=False):
+                 return_region=False,
+                 patch_left=0, patch_right=0):
         
         self.dataframe = dataframe        
         self.genome_fasta = genome_fasta
@@ -122,6 +123,8 @@ class _SeqChromDatasetByDataFrame(IterableDataset):
 
         self.transforms = transforms
         self.return_region = return_region
+        self.patch_left = patch_left
+        self.patch_right = patch_right
         
         self.start = 0; self.end = len(self.dataframe)
     
@@ -157,7 +160,9 @@ class _SeqChromDatasetByDataFrame(IterableDataset):
                     bigwigs=self.bigwigs,
                     target=self.target_pysam,
                     strand=item.strand,
-                    transforms=self.transforms
+                    transforms=self.transforms,
+                    patch_left=self.patch_left,
+                    patch_right=self.patch_right
                 )
             except utils.BigWigInaccessible as e:
                 logging.warn(f"Inaccessible bigwig error detected in region {item.chrom}:{item.start}-{item.end}, Skipping...")
@@ -187,14 +192,17 @@ class _SeqChromDatasetByBed(_SeqChromDatasetByDataFrame):
     :param transforms: A dictionary of functions to transform the output data, accepted keys are *["seq", "chrom", "target", "label"]*
     :type transforms: dict of functions
     """
-    def __init__(self, bed: str, genome_fasta: str, bigwig_filelist:list, target_bam=None, transforms:dict=None, return_region=False):
+    def __init__(self, bed: str, genome_fasta: str, bigwig_filelist:list, target_bam=None, 
+                 transforms:dict=None, return_region=False,
+                 patch_left=0, patch_right=0):
         dataframe = pd.read_table(bed, header=None, names=['chrom', 'start', 'end', 'label', 'score', 'strand' ])
         super().__init__(dataframe,
                          genome_fasta,
                          bigwig_filelist,
                          target_bam,
                          transforms,
-                         return_region)
+                         return_region,
+                         patch_left, patch_right)
 
 SeqChromDatasetByBed = seqChromLoaderCurry(_SeqChromDatasetByBed)
 

{seqchromloader-0.7.3 → seqchromloader-0.7.4}/seqchromloader/writer.py

@@ -186,7 +186,7 @@ def dump_data_webdataset_worker(coords,
         
         if batch_size is None:    
             feature_dict = defaultdict()
-            feature_dict["__key__"] = f"{rindex}_{item.chrom}:{item.start}-{item.end}_{item.strand}" 
+            feature_dict["__key__"] = f"{rindex}_{item.chrom}:{item.start-patch_left}-{item.end+patch_right}_{item.strand}" 
             feature_dict["seq.npy"] = feature['seq']
             feature_dict["chrom.npy"] = feature['chrom']
             feature_dict["target.npy"] = feature['target']

{seqchromloader-0.7.3 → seqchromloader-0.7.4}/seqchromloader.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: seqchromloader
-Version: 0.7.3
+Version: 0.7.4
 Summary: Sequence and chromatin dataloader for deep learning
 Home-page: https://github.com/yztxwd/seqchromloader
 Author-email: yztxwd@gmail.com

{seqchromloader-0.7.3 → seqchromloader-0.7.4}/setup.py

@@ -20,7 +20,7 @@ setup(
     # eg: 1.0.0, 1.0.1, 3.0.2, 5.0-beta, etc.
     # You CANNOT upload two versions of your package with the same version number
     # This field is REQUIRED
-    version="0.7.3",
+    version="0.7.4",
 
     # The packages that constitute your project.
     # For my project, I have only one - "pydash".