sai-pg 1.0.0__py3-none-any.whl → 1.0.1__py3-none-any.whl

sai/utils/generators/chunk_generator.py

@@ -131,18 +131,12 @@ class ChunkGenerator(DataGenerator):
         remainder = len(windows) % num_chunks
         result = []
         start_idx = 0
-        prev_end = None
 
         for i in range(num_chunks):
             end_idx = start_idx + avg + (1 if i < remainder else 0)
             sub = windows[start_idx:end_idx]
             if sub:
-                min_start = sub[0][0]
-                max_end = sub[-1][1]
-                if (prev_end is not None) and (min_start < prev_end):
-                    min_start = prev_end + 1
-                result.append((min_start, max_end))
-                prev_end = max_end
+                result.append((sub[0][0], sub[-1][1]))
             start_idx = end_idx
 
         return result

sai/utils/generators/window_generator.py

@@ -125,6 +125,7 @@ class WindowGenerator(DataGenerator):
                 ),
                 window_size=self.win_len,
                 step_size=self.win_step,
+                start=start,
             )
             for tgt_pop in self.tgt_samples
         }
@@ -152,21 +153,21 @@ class WindowGenerator(DataGenerator):
             for start, end in self.tgt_windows[tgt_pop]:
                 ref_gts = self.ref_data[ref_pop].GT[
                     (self.ref_data[ref_pop].POS >= start)
-                    & (self.ref_data[ref_pop].POS < end)
+                    & (self.ref_data[ref_pop].POS <= end)
                 ]
                 tgt_gts = self.tgt_data[tgt_pop].GT[
                     (self.tgt_data[tgt_pop].POS >= start)
-                    & (self.tgt_data[tgt_pop].POS < end)
+                    & (self.tgt_data[tgt_pop].POS <= end)
                 ]
                 src_gts_list = [
                     self.src_data[src_pop].GT[
                         (self.src_data[src_pop].POS >= start)
-                        & (self.src_data[src_pop].POS < end)
+                        & (self.src_data[src_pop].POS <= end)
                     ]
                     for src_pop in src_comb
                 ]
 
-                sub_pos = tgt_pos[(tgt_pos >= start) & (tgt_pos < end)]
+                sub_pos = tgt_pos[(tgt_pos >= start) & (tgt_pos <= end)]
 
                 yield {
                     "chr_name": self.chr_name,

sai/utils/utils.py

@@ -608,6 +608,7 @@ def split_genome(
     pos: np.ndarray,
     window_size: int,
     step_size: int,
+    start: int = None,
 ) -> list[tuple]:
     """
     Creates sliding windows along the genome based on variant positions.
@@ -620,6 +621,9 @@ def split_genome(
         Length of each sliding window.
     step_size : int
         Step size of the sliding windows.
+    start: int, optional
+        Minimum starting coordinate for the first window. The first window will start
+        no smaller than this value. Default is None.
 
     Returns
     -------
@@ -644,7 +648,9 @@ def split_genome(
 
     window_positions = []
     win_start = (pos[0] + step_size) // step_size * step_size - window_size + 1
-    win_start = max(win_start, 1)
+    if start is None:
+        start = 1
+    win_start = max(win_start, start)
 
     # Create windows based on step size and window size
     while win_start <= pos[-1]:

{sai_pg-1.0.0.dist-info → sai_pg-1.0.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sai-pg
-Version: 1.0.0
+Version: 1.0.1
 Summary: A Python Package for Statistics for Adaptive Introgression
 Home-page: https://github.com/xin-huang/sai
 Author: Xin Huang

{sai_pg-1.0.0.dist-info → sai_pg-1.0.1.dist-info}/RECORD

@@ -10,11 +10,11 @@ sai/stats/__init__.py,sha256=ZLPiBk86c9R8ZFx9y5VF7Up2v4JS8WsEMfsVNlj7nXY,724
 sai/stats/features.py,sha256=7CI9i3csVH9sXryWGB0uLSJB6BbM9gJtA1IPN2EeT0k,11844
 sai/utils/__init__.py,sha256=B3ZcC1ALSWieGHPiqXKBFQRTrnlTX4TaHc3tCx9fj0w,782
 sai/utils/genomic_dataclasses.py,sha256=HBYp2dehdW_y3Pd6Un8XFMnN1Odg1EiZb9ci1syIibU,1443
-sai/utils/utils.py,sha256=Nnb1tPHWk9vmlRmUJ65_FEmLfDEsT41HZ9WvJ0ps7Vs,23127
+sai/utils/utils.py,sha256=omWRgtc_n4FlbANbkg-Zx1hRoNgXNn14GuAZq4gD7dI,23358
 sai/utils/generators/__init__.py,sha256=hpE4PUQIOZQXzdpSx7dEllecDoOfxIWXNu1-WHa_VcM,858
-sai/utils/generators/chunk_generator.py,sha256=Jh0wDOR6Z5PXlzOTncmY-pr9PotnmfiWTZQL2aaVGAs,4565
+sai/utils/generators/chunk_generator.py,sha256=1qFTcduHIaRwxuEhxRNlCRbGJbPzN3A0waMjTi52UII,4319
 sai/utils/generators/data_generator.py,sha256=bVz9KPjJSL4becTsVZ-zH4i40y5UDA0JJOtzbwf5n84,1635
-sai/utils/generators/window_generator.py,sha256=G8aMmJbB8uWLNaoJAGSeBFTCvAib7DdWvbXDk9JvDdM,9066
+sai/utils/generators/window_generator.py,sha256=-3FJtdBaHMiGk85e_vyPWDYkbdCfqS2IdD_hTgpgMZg,9099
 sai/utils/multiprocessing/__init__.py,sha256=mI_iVjOJschKjIcneo99utCMVjR0K3UpEjA_PMcmLm4,790
 sai/utils/multiprocessing/mp_manager.py,sha256=xbgje4Fi4U9WD7eqMlUkTLKLQ9aNV_hIG7nTYvniGFk,10263
 sai/utils/multiprocessing/mp_pool.py,sha256=wqgp7E7fU2MXKsv7d8kpIp40SLjqS8voxbB_FkVwczA,2273
@@ -22,9 +22,9 @@ sai/utils/preprocessors/__init__.py,sha256=q25uXnq4jTrIVBbl1HFv3ZgEV0d79fCMp4tTZ
 sai/utils/preprocessors/chunk_preprocessor.py,sha256=mNmc22PmIRZZjYoyktxKSRGSGzN_e1Cyi392VxTbtWc,5204
 sai/utils/preprocessors/data_preprocessor.py,sha256=XSQN_kXUfyCtIpqIf7TcqWIPcSk8kynHPH-KY9gh5VY,3595
 sai/utils/preprocessors/feature_preprocessor.py,sha256=Kmc7VsO9YJcbFwWk-Jha5Rzq7gZou1pwYh93Wpd9r5o,7143
-sai_pg-1.0.0.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-sai_pg-1.0.0.dist-info/METADATA,sha256=e16SXHsIZBXK5EtJDbDhyPJdEXcKes88sCNPls_s3y4,1789
-sai_pg-1.0.0.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-sai_pg-1.0.0.dist-info/entry_points.txt,sha256=uK34frE2UhEDNa5ISbGOtjF2HGAXF5uG_EgMocesEPs,42
-sai_pg-1.0.0.dist-info/top_level.txt,sha256=Kvf-0z0sZYQbLuM6ta183EOzD37k3VZbRH-h7HXntR0,4
-sai_pg-1.0.0.dist-info/RECORD,,
+sai_pg-1.0.1.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+sai_pg-1.0.1.dist-info/METADATA,sha256=zHMDzO-qsymZICEJSGn0H8o14qkrIAFoCim5sBhHlq0,1789
+sai_pg-1.0.1.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+sai_pg-1.0.1.dist-info/entry_points.txt,sha256=uK34frE2UhEDNa5ISbGOtjF2HGAXF5uG_EgMocesEPs,42
+sai_pg-1.0.1.dist-info/top_level.txt,sha256=Kvf-0z0sZYQbLuM6ta183EOzD37k3VZbRH-h7HXntR0,4
+sai_pg-1.0.1.dist-info/RECORD,,