Mouse-AAV-OSN

Computational analysis for "Comparative Analysis of AAV Serotypes for the Transduction of Olfactory Sensory Neurons" by Belfort and Jia et al. 2024

GEO Submission Scripts Documentation: src/snrna/utils

This document describes three Python scripts designed to handle AnnData objects for Gene Expression Omnibus (GEO) submissions: a preprocessing script, an AnnData to GEO conversion script, and a GEO to AnnData reconstruction script.

Requirements

Python pandas anndata scipy numpy

1. Preprocessing Script: preprocess_geo_files.py

Purpose

Standardizes the file structure of GEO submission files by removing prefixes and organizing files into a consistent format.

Input

• input_dir: Path to the directory containing all screen folders and final object folders.

Output

• Standardized file structure in the specified output directory.

Functionality

• Processes each subdirectory in the input directory.
• Removes prefixes from filenames.
• Copies files to a new directory structure with standardized names.

Usage

python preprocess_geo_files.py /path/to/input/directory /path/to/output/directory

2. AnnData to GEO Conversion Script: anndata_to_geo.py

Purpose

Converts an AnnData object to the format required for GEO submission.

Input

• input_file: Path to the input .h5ad file.
• output_dir: Path to the output directory for GEO submission files.
• --prefix (optional): Prefix to add to output files.
• --chunk_size (optional): Chunk size for splitting large matrices (default: 1,000,000).

Output

• GEO submission files in the specified output directory, including:
  • Cell metadata (obs)
  • Gene metadata (var)
  • Count matrix (in chunks)
  • Layer data (if present)
  • Raw data (if present)
  • README file

Functionality

• Reads an AnnData object.
• Saves metadata, count matrix, layers, and raw data (if present) in GEO-compatible format.
• Splits large matrices into chunks.
• Generates a README file describing the dataset.

Usage

python anndata_to_geo.py input.h5ad /path/to/output/directory --prefix optional_prefix_ --chunk_size 500000

3. GEO to AnnData Reconstruction Script: geo_to_anndata.py

Purpose

Reconstructs an AnnData object from GEO submission files.

Input

• input_dir: Path to the directory containing GEO submission files.
• output_file: Path to save the reconstructed .h5ad file.

Output

• Reconstructed AnnData object saved as an .h5ad file.

Functionality

• Automatically detects file prefixes.
• Reads cell metadata, gene metadata, and count matrix.
• Reconstructs layers and raw data if present.
• Converts matrices to sparse format if they contain more than 50% zero values.
• Creates and saves an AnnData object.

Usage

python geo_to_anndata.py /path/to/geo/submission/files /path/to/output/reconstructed_file.h5ad

Workflow Example

1. Standardize your file structure:

   python preprocess_geo_files.py /original/files /standardized/files

2. Convert AnnData to GEO format (if needed):

   python anndata_to_geo.py input.h5ad /geo/submission/files --prefix screen1_

3. Reconstruct AnnData from GEO files:

   python geo_to_anndata.py /geo/submission/files reconstructed_data.h5ad

Remember to adjust file paths and options according to your specific dataset and requirements.