This is a Python package that downloads and preprocesses the LARRY dataset from the AllonKleinLab GitHub repository. The LARRY dataset is a group of single-cell lineage-traced datasets that have been used to study transcriptional landscapes and cell fate during differentiation [1]. There are three datasets within this group of datasets, all using hematopoietic progenitor cells from mouse bone marrow and the LARRY lentiviral barcoding strategy.

1. in vitro
2. in vivo - transplanted into mice
3. Cytokine-perturbed (in vitro) - split between various cytokine culture conditions.

The package includes functions to download the LARRY dataset, format the files into AnnData and perform preprocessing / gene-filtering steps. The preprocessed data can then be split into test and training sets for use in different machine learning tasks. Dimension reduction can be performed after data splitting to respect information leakage or on all of the data.

Installation

pip distribution

pip install larry-dataset

Development version

git clone https://github.com/mvinyard/LARRY-dataset.git; cd LARRY-dataset

pip install -e .

Quickstart

Downloads pre-processed data from AllonKleinLab/paper-data to ./KleinLabData (by default). The data is formatted into AnnData and returned to the user. A .h5ad file is also saved, locally. The data downloading and conversion step take several minutes due to the large expression normed_counts matrices though this only happens once.

import larry
    
dataset = "in_vitro" # can also choose from: "in_vivo" or "cytokine_perturbation"
adata = larry.fetch(dataset)

AnnData object with n_obs × n_vars = 130887 × 25289
    obs: 'Library', 'Cell barcode', 'Time point', 'Starting population', 'Cell type annotation', 'Well', 'SPRING-x', 'SPRING-y'
    var: 'gene_name'
    obsm: 'X_clone'

import larry

LARRY_LightningData = larry.LARRY_LightningDataModule()
LARRY_LightningData.prepare_data()

 AnnData object with n_obs × n_vars = 130887 × 25289
    obs: 'Library', 'Cell barcode', 'Time point', 'Starting population', 'Cell type annotation', 'Well', 'SPRING-x', 'SPRING-y'
    var: 'gene_name'
    uns: 'dataset', 'h5ad_path'
    obsm: 'X_clone'
Preprocessing performed previously. Loading...done.

Under the hood, the LARRY_LightningData calls larry.fetch() and larry.pp.Yeo2021_recipe(), and if task == "fate_prediction", larry.pp.annotate_fate_test_train()

LARRY_LightningData.adata

Print the updated adata:

AnnData object with n_obs × n_vars = 130887 × 25289
    obs: 'Library', 'Cell barcode', 'Time point', 'Starting population', 'Cell type annotation', 'Well', 'SPRING-x', 'SPRING-y', 'cell_idx', 'clone_idx'
    var: 'gene_name', 'highly_variable', 'corr_cell_cycle', 'pass_filter'
    uns: 'dataset', 'h5ad_path', 'highly_variable_genes_idx', 'n_corr_cell_cycle', 'n_hv', 'n_mito', 'n_pass', 'n_total', 'pp_h5ad_path'
    obsm: 'X_clone', 'X_pca', 'X_scaled', 'X_umap'

Sources

Repositories

• AllonKleinLab/paper-data
• AllonKleinLab/LARRY

References

1. Weinreb, C., Rodriguez-Fraticelli, A., Camargo, F.D., Klein, A.M. Lineage tracing on transcriptional landscapes links state to fate during differentiation. Science 80 (2020). https://doi.org/10.1126/science.aaw3381

---

Please email Michael E. Vinyard (mvinyard@broadinstitute.org) with any questions or interests.