scdataloader 1.9.1__py3-none-any.whl → 1.9.2__py3-none-any.whl

scdataloader/VERSION

@@ -1 +1 @@
-1.9.1
+1.9.2

scdataloader/__init__.py

@@ -1,7 +1,8 @@
+from importlib.metadata import version
+
 from .collator import Collator
 from .data import Dataset, SimpleAnnDataset
 from .datamodule import DataModule
 from .preprocess import Preprocessor
-from importlib.metadata import version
 
 __version__ = version("scdataloader")

scdataloader/collator.py

@@ -148,7 +148,6 @@ class Collator:
                         :, self.accepted_genes[organism_id]
                     ]
             if self.how == "most expr":
-                nnz_loc = np.where(expr > 0)[0]
                 if "knn_cells" in elem:
                     nnz_loc = np.where(expr + elem["knn_cells"].sum(0) > 0)[0]
                     ma = self.max_len if self.max_len < len(nnz_loc) else len(nnz_loc)
@@ -161,14 +160,18 @@ class Collator:
                 # loc = np.argsort(expr)[-(self.max_len) :][::-1]
             elif self.how == "random expr":
                 nnz_loc = np.where(expr > 0)[0]
-                loc = nnz_loc[
-                    np.random.choice(
-                        len(nnz_loc),
-                        self.max_len if self.max_len < len(nnz_loc) else len(nnz_loc),
-                        replace=False,
-                        # p=(expr.max() + (expr[nnz_loc])*19) / expr.max(), # 20 at most times more likely to be selected
-                    )
-                ]
+                loc = (
+                    nnz_loc[
+                        np.random.choice(
+                            len(nnz_loc),
+                            self.max_len,
+                            replace=False,
+                            # p=(expr.max() + (expr[nnz_loc])*19) / expr.max(), # 20 at most times more likely to be selected
+                        )
+                    ]
+                    if self.max_len < len(nnz_loc)
+                    else nnz_loc
+                )
             elif self.how in ["all", "some"]:
                 loc = np.arange(len(expr))
             else:
@@ -179,23 +182,19 @@ class Collator:
                 "all",
                 "some",
             ]:
+                ma = self.add_zero_genes + (
+                    0 if self.max_len < len(nnz_loc) else self.max_len - len(nnz_loc)
+                )
                 if "knn_cells" in elem:
                     # we complete with genes expressed in the knn
-                    nnz_loc = np.where(elem["knn_cells"].sum(0) > 0)[0]
-                    ma = self.max_len if self.max_len < len(nnz_loc) else len(nnz_loc)
                     # which is not a zero_loc in this context
-                    zero_loc = np.argsort(elem["knn_cells"].sum(0))[-(ma):][::-1]
+                    zero_loc = np.argsort(elem["knn_cells"].sum(0))[-ma:][::-1]
                 else:
                     zero_loc = np.where(expr == 0)[0]
                     zero_loc = zero_loc[
                         np.random.choice(
                             len(zero_loc),
-                            self.add_zero_genes
-                            + (
-                                0
-                                if self.max_len < len(nnz_loc)
-                                else self.max_len - len(nnz_loc)
-                            ),
+                            ma,
                             replace=False,
                         )
                     ]

scdataloader/datamodule.py

@@ -16,7 +16,7 @@ from torch.utils.data.sampler import (
 
 from .collator import Collator
 from .data import Dataset
-from .utils import getBiomartTable, slurm_restart_count
+from .utils import getBiomartTable
 
 FILE_DIR = os.path.dirname(os.path.abspath(__file__))
 
@@ -28,7 +28,7 @@ class DataModule(L.LightningDataModule):
         clss_to_weight: list = ["organism_ontology_term_id"],
         organisms: list = ["NCBITaxon:9606"],
         weight_scaler: int = 10,
-        train_oversampling_per_epoch: float = 0.1,
+        n_samples_per_epoch: int = 2_000_000,
         validation_split: float = 0.2,
         test_split: float = 0,
         gene_embeddings: str = "",
@@ -53,7 +53,6 @@ class DataModule(L.LightningDataModule):
         ],
         metacell_mode: float = 0.0,
         get_knn_cells: bool = False,
-        modify_seed_on_requeue: bool = True,
         **kwargs,
     ):
         """
@@ -67,7 +66,7 @@ class DataModule(L.LightningDataModule):
             collection_name (str): The lamindb collection to be used.
             organisms (list, optional): The organisms to include in the dataset. Defaults to ["NCBITaxon:9606"].
             weight_scaler (int, optional): how much more you will see the most present vs less present category.
-            train_oversampling_per_epoch (float, optional): The proportion of the dataset to include in the training set for each epoch. Defaults to 0.1.
+            n_samples_per_epoch (int, optional): The number of samples to include in the training set for each epoch. Defaults to 2_000_000.
             validation_split (float, optional): The proportion of the dataset to include in the validation split. Defaults to 0.2.
             test_split (float, optional): The proportion of the dataset to include in the test split. Defaults to 0.
                 it will use a full dataset and will round to the nearest dataset's cell count.
@@ -88,7 +87,6 @@ class DataModule(L.LightningDataModule):
             hierarchical_clss (list, optional): List of hierarchical classes. Defaults to [].
             metacell_mode (float, optional): The probability of using metacell mode. Defaults to 0.0.
             clss_to_predict (list, optional): List of classes to predict. Defaults to ["organism_ontology_term_id"].
-            modify_seed_on_requeue (bool, optional): Whether to modify the seed on requeue. Defaults to True.
             get_knn_cells (bool, optional): Whether to get the k-nearest neighbors of each queried cells. Defaults to False.
             **kwargs: Additional keyword arguments passed to the pytorch DataLoader.
             see @file data.py and @file collator.py for more details about some of the parameters
@@ -171,13 +169,11 @@ class DataModule(L.LightningDataModule):
         self.assays_to_drop = assays_to_drop
         self.n_samples = len(mdataset)
         self.weight_scaler = weight_scaler
-        self.train_oversampling_per_epoch = train_oversampling_per_epoch
+        self.n_samples_per_epoch = n_samples_per_epoch
         self.clss_to_weight = clss_to_weight
         self.train_weights = None
         self.train_labels = None
-        self.modify_seed_on_requeue = modify_seed_on_requeue
         self.nnz = None
-        self.restart_num = 0
         self.test_datasets = []
         self.test_idx = []
         super().__init__()
@@ -190,7 +186,7 @@ class DataModule(L.LightningDataModule):
             f"\ttest_split={self.test_split},\n"
             f"\tn_samples={self.n_samples},\n"
             f"\tweight_scaler={self.weight_scaler},\n"
-            f"\ttrain_oversampling_per_epoch={self.train_oversampling_per_epoch},\n"
+            f"\tn_samples_per_epoch={self.n_samples_per_epoch},\n"
             f"\tassays_to_drop={self.assays_to_drop},\n"
             f"\tnum_datasets={len(self.dataset.mapped_dataset.storages)},\n"
             f"\ttest datasets={str(self.test_datasets)},\n"
@@ -336,18 +332,16 @@ class DataModule(L.LightningDataModule):
     def train_dataloader(self, **kwargs):
         # train_sampler = WeightedRandomSampler(
         #    self.train_weights[self.train_labels],
-        #    int(self.n_samples*self.train_oversampling_per_epoch),
+        #    int(self.n_samples*self.n_samples_per_epoch),
         #    replacement=True,
         # )
         try:
             train_sampler = LabelWeightedSampler(
                 label_weights=self.train_weights,
                 labels=self.train_labels,
-                num_samples=int(self.n_samples * self.train_oversampling_per_epoch),
+                num_samples=int(self.n_samples_per_epoch),
                 element_weights=self.nnz,
                 replacement=self.replacement,
-                restart_num=self.restart_num,
-                modify_seed_on_requeue=self.modify_seed_on_requeue,
             )
         except ValueError as e:
             raise ValueError(e + "have you run `datamodule.setup()`?")
@@ -397,8 +391,6 @@ class LabelWeightedSampler(Sampler[int]):
     num_samples: int
     nnz: Optional[Sequence[int]]
     replacement: bool
-    restart_num: int
-    modify_seed_on_requeue: bool
     # when we use, just set weights for each classes(here is: np.ones(num_classes)), and labels of a dataset.
     # this will result a class-balanced sampling, no matter how imbalance the labels are.
 
@@ -409,15 +401,12 @@ class LabelWeightedSampler(Sampler[int]):
         num_samples: int,
         replacement: bool = True,
         element_weights: Sequence[float] = None,
-        restart_num: int = 0,
-        modify_seed_on_requeue: bool = True,
     ) -> None:
         """
 
         :param label_weights: list(len=num_classes)[float], weights for each class.
         :param labels: list(len=dataset_len)[int], labels of a dataset.
         :param num_samples: number of samples.
-        :param restart_num: if we are continuing a previous run, we need to restart the sampler from the same point.
         """
 
         super(LabelWeightedSampler, self).__init__(None)
@@ -433,8 +422,6 @@ class LabelWeightedSampler(Sampler[int]):
         )
         self.replacement = replacement
         self.num_samples = num_samples
-        self.restart_num = slurm_restart_count(use_mine=True) + restart_num
-        self.modify_seed_on_requeue = modify_seed_on_requeue
         # list of tensor.
         self.klass_indices = [
             (self.labels == i_klass).nonzero().squeeze(1)
@@ -447,9 +434,6 @@ class LabelWeightedSampler(Sampler[int]):
             self.label_weights,
             num_samples=self.num_samples,
             replacement=True,
-            generator=None
-            if self.restart_num == 0 and not self.modify_seed_on_requeue
-            else torch.Generator().manual_seed(self.restart_num),
         )
         sample_indices = torch.empty_like(sample_labels)
         for i_klass, klass_index in enumerate(self.klass_indices):
@@ -465,17 +449,11 @@ class LabelWeightedSampler(Sampler[int]):
                     if not self.replacement and len(klass_index) < len(left_inds)
                     else len(left_inds),
                     replacement=self.replacement,
-                    generator=None
-                    if self.restart_num == 0 and not self.modify_seed_on_requeue
-                    else torch.Generator().manual_seed(self.restart_num),
                 )
             elif self.replacement:
                 right_inds = torch.randint(
                     len(klass_index),
                     size=(len(left_inds),),
-                    generator=None
-                    if self.restart_num == 0 and not self.modify_seed_on_requeue
-                    else torch.Generator().manual_seed(self.restart_num),
                 )
             else:
                 maxelem = (
@@ -485,6 +463,7 @@ class LabelWeightedSampler(Sampler[int]):
                 )
                 right_inds = torch.randperm(len(klass_index))[:maxelem]
             sample_indices[left_inds[: len(right_inds)]] = klass_index[right_inds]
+            # if there are more left_inds than right_inds, we need to drop the extra ones
             if len(right_inds) < len(left_inds):
                 sample_indices[left_inds[len(right_inds) :]] = -1
         # drop all -1
@@ -492,7 +471,6 @@ class LabelWeightedSampler(Sampler[int]):
         # torch shuffle
         sample_indices = sample_indices[torch.randperm(len(sample_indices))]
         self.num_samples = len(sample_indices)
-        # raise Exception("stop")
         yield from iter(sample_indices.tolist())
 
     def __len__(self):

scdataloader/mapped.py

@@ -1,3 +1,10 @@
+# Portions of this file are derived from Lamin Labs
+# Copyright 2024 Lamin Labs
+# Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+# The rest of this file is licensed under MIT
+# Please see https://github.com/laminlabs/lamindb/blob/main/lamindb/core/_mapped_collection.py
+# for the original implementation
+
 from __future__ import annotations
 
 from collections import Counter

scdataloader/preprocess.py

@@ -1,3 +1,4 @@
+import gc
 from typing import Callable, Optional, Union
 from uuid import uuid4
 
@@ -9,7 +10,7 @@ import scanpy as sc
 from anndata import AnnData, read_h5ad
 from scipy.sparse import csr_matrix
 from upath import UPath
-import gc
+
 from scdataloader import utils as data_utils
 
 FULL_LENGTH_ASSAYS = [
@@ -270,10 +271,12 @@ class Preprocessor:
             var = ens_var
 
         adata = adata[:, var.index]
-        var = var.sort_values(by="ensembl_gene_id").set_index("ensembl_gene_id")
+        #        var = var.sort_values(by="ensembl_gene_id").set_index("ensembl_gene_id")
         # Update adata with combined genes
-        adata.var = var
-        genesdf = genesdf.set_index("ensembl_gene_id")
+        if "ensembl_gene_id" in var.columns:
+            adata.var = var.set_index("ensembl_gene_id")
+        else:
+            adata.var = var
         # Drop duplicate genes, keeping first occurrence
         adata = adata[:, ~adata.var.index.duplicated(keep="first")]
 
@@ -503,7 +506,7 @@ class LaminPreprocessor(Preprocessor):
                     continue
                 print(file)
 
-                path = cache_path(file) if self.force_preloaded else file.cache()
+                _ = cache_path(file) if self.force_preloaded else file.cache()
                 backed = file.open()
                 # backed = read_h5ad(path, backed="r")
                 if "is_primary_data" in backed.obs.columns:

{scdataloader-1.9.1.dist-info → scdataloader-1.9.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scdataloader
-Version: 1.9.1
+Version: 1.9.2
 Summary: a dataloader for single cell data in lamindb
 Project-URL: repository, https://github.com/jkobject/scDataLoader
 Author-email: jkobject <jkobject@gmail.com>
@@ -15,7 +15,7 @@ Requires-Dist: django>=4.0.0
 Requires-Dist: harmonypy>=0.0.10
 Requires-Dist: ipykernel>=6.20.0
 Requires-Dist: jupytext>=1.16.0
-Requires-Dist: lamindb[bionty,cellregistry,jupyter,ourprojects,zarr]<2,>=1.0.4
+Requires-Dist: lamindb[bionty,cellregistry,jupyter,zarr]<2,>=1.0.4
 Requires-Dist: leidenalg>=0.8.0
 Requires-Dist: matplotlib>=3.5.0
 Requires-Dist: numpy==1.26.0
@@ -71,7 +71,16 @@ It allows you to:
 3. create a more complex single cell dataset
 4. extend it to your need
 
-built on top of `lamindb` and the `.mapped()` function by Sergei: https://github.com/Koncopd 
+built on top of `lamindb` and the `.mapped()` function by Sergei: https://github.com/Koncopd
+
+```
+Portions of the mapped.py file are derived from Lamin Labs
+Copyright 2024 Lamin Labs
+Licensed under the Apache License, Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
+The rest of the package is licensed under MIT License, see LICENSE for details
+Please see https://github.com/laminlabs/lamindb/blob/main/lamindb/core/_mapped_collection.py
+for the original implementation
+```
 
 The package has been designed together with the [scPRINT paper](https://doi.org/10.1101/2024.07.29.605556) and [model](https://github.com/cantinilab/scPRINT).
 

scdataloader-1.9.2.dist-info/RECORD

@@ -0,0 +1,16 @@
+scdataloader/VERSION,sha256=u1mRkBjudkgsI_HORCGc5MnGOTxl7w3a5A0y151BO7U,6
+scdataloader/__init__.py,sha256=Z5HURehoWw1GrecImmTXIkv4ih8Q5RxNQWPm8zjjXOA,226
+scdataloader/__main__.py,sha256=3aZnqYrH8XDT9nW9Dbb3o9kr-sx1STmXDQHxBo_h_q0,8719
+scdataloader/base.py,sha256=M1gD59OffRdLOgS1vHKygOomUoAMuzjpRtAfM3SBKF8,338
+scdataloader/collator.py,sha256=qb1SDQ358R4w56cxOXvVLpodlZpGfVDCXocFIhqpJ0I,12867
+scdataloader/config.py,sha256=YQUKCyYTg4wTseBWumPDHKtmqI7DMR-zu5FPJUWkG-c,6549
+scdataloader/data.py,sha256=xWlNU6cJmrzP4BFMsJDIksLaxe1pUfgDBlQ_IeLIXj0,15578
+scdataloader/datamodule.py,sha256=MaBSH0MqZdDbwiGnCM4xjz0KF05WT00sycJStS7GL5w,19786
+scdataloader/mapped.py,sha256=qzhGYQ2S3IDXnnO1EM1wO_png5lDiOtfuDeYc1pQaXg,27303
+scdataloader/preprocess.py,sha256=LNXlP80rj8Ze2ElyIgLuF9x_lNA78jAI-seGMOyMKGs,37496
+scdataloader/utils.py,sha256=7ycZoV01Gn3WDHOTmXqxMXlzBPSfYtjc9NbGI7gjdwI,28445
+scdataloader-1.9.2.dist-info/METADATA,sha256=jLx2v-SJTAv-jsUjmgLNIrTenwtnfUHTnKYuy9hI3lg,10363
+scdataloader-1.9.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+scdataloader-1.9.2.dist-info/entry_points.txt,sha256=VXAN1m_CjbdLJ6SKYR0sBLGDV4wvv31ri7fWWuwbpno,60
+scdataloader-1.9.2.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+scdataloader-1.9.2.dist-info/RECORD,,

scdataloader-1.9.1.dist-info/RECORD

@@ -1,16 +0,0 @@
-scdataloader/VERSION,sha256=CrT6GQyhsdH2MKyHzzRnK_uvE5XQFIx1aNOOWDNbjWg,6
-scdataloader/__init__.py,sha256=1SyT5MzcFl8mfp5NB4idgYQ4insXbDRd-EBNvoz_dXQ,225
-scdataloader/__main__.py,sha256=3aZnqYrH8XDT9nW9Dbb3o9kr-sx1STmXDQHxBo_h_q0,8719
-scdataloader/base.py,sha256=M1gD59OffRdLOgS1vHKygOomUoAMuzjpRtAfM3SBKF8,338
-scdataloader/collator.py,sha256=nznnyb6H_OYgBbbbbs1ZBeYDnlrtJ6jAgJeoSf7CUXQ,13065
-scdataloader/config.py,sha256=YQUKCyYTg4wTseBWumPDHKtmqI7DMR-zu5FPJUWkG-c,6549
-scdataloader/data.py,sha256=xWlNU6cJmrzP4BFMsJDIksLaxe1pUfgDBlQ_IeLIXj0,15578
-scdataloader/datamodule.py,sha256=6Oby-BySXaWYr34PocgCq25FLH1QUX-EsWOZI6EVjgw,21128
-scdataloader/mapped.py,sha256=RHbLCZtjWTlNX_7Iyyw6g53pZ2_5R6QU7gRJMjMxGKg,26953
-scdataloader/preprocess.py,sha256=50Gr1ZeuxhHd-0IGm1UbGmMpOf1UAgmKmRjPCgdrOwE,37424
-scdataloader/utils.py,sha256=7ycZoV01Gn3WDHOTmXqxMXlzBPSfYtjc9NbGI7gjdwI,28445
-scdataloader-1.9.1.dist-info/METADATA,sha256=NjiK3ZWmqLvqfs_1jt5D2eTH8p1gLWPWhZeuxl-cRJg,9986
-scdataloader-1.9.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-scdataloader-1.9.1.dist-info/entry_points.txt,sha256=VXAN1m_CjbdLJ6SKYR0sBLGDV4wvv31ri7fWWuwbpno,60
-scdataloader-1.9.1.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-scdataloader-1.9.1.dist-info/RECORD,,