Skip to content

Latest commit

 

History

History
139 lines (98 loc) · 8.19 KB

README.md

File metadata and controls

139 lines (98 loc) · 8.19 KB
Raw

Cell Line Metadata Database

This repository contains scripts for creating and managing cell line metadata database to enable the annotation of SDRFs for cell lines datasets. The main driver use case is the annotation of SDRF datasets for the quantms.org resource. This repo uses multiple ontologies and natural language processing (NLP) to annotate cell lines in SDRF files.

Motivation

Cell lines are a fundamental part of biological research, and they are used in a wide range of experiments. However, cell line metadata can be inconsistent and difficult to manage. Here we are creating a DB that can be used to annotate/validate proteomics SDRF for cell lines studies. These are the major sources of cell line metadata:

  • CelloSaurus: CelloSaurus is the main source used in our database. The source of the metadata can be downloaded from cellosaurus.txt. We converted the file to a shorter version with only the fields that we are interested and the taxonomy. We use the script pycls cellosaurus-database to create the database.
  • Cell model passports: The cell model passports are a collection of cell lines from multiple sources. We use the file model_list_20240110.csv to create a database extracting only the cell lines information pycls cell-passports-to-database.
  • EA: Expression Atlas has been curating for more than 10 years the metadata of multiple RNA experiments. We collect multiple cell lines experiments from EA in folder ea; and try to create a catalog of cell lines metadata as an extra source.
  • MONDO: The Monarch Disease Ontology (MONDO) is used to annotate the disease of the cell line.
  • BTO: The BRENDA Tissue Ontology (BTO) is used to annotate an extra reference for the cell line ID.

Note: Additionally, we use other resources such as Coriell cell line Catalog, cell bank riken and atcc for manual annotation of cell lines in the database.

Features for every cell line

The database is created in the following path cl-annotations-db.tsv and contains the following fields:

  • cell line: The cell line name as defined by the curation team (ai or manual).
  • cellosaurus name: The cell line name as annotated in Cellosaurus ID
  • cellosaurus accession: The cell line accession as annotated in Cellsaurus AC
  • bto cell line: The cell line name as annotated in BTO
  • organism: The organism of the cell line as annotated in Cellosaurus
  • organism part: This information is not available in Cellosaurus, we use other sources to annotate this field.
  • sampling site: The sampling site of the cell line as annotated in Cellosaurus. If the information is not available, we use other sources to annotate this field.
  • age: The age of the cell line as annotated in Cellosaurus. If the age is not available (empty), we annotated the age from other sources such as atcc or Coriell cell line Catalog
  • developmental stage: The developmental stage of the cell line as annotated in Cellosaurus; if the information is not available is inferred from the age of the cell line.
  • sex: Sex as provided by Cellosaurus
  • ancestry category: The ancestry category of the cell line as annotated in Cellosaurus. If not available we use other sources.
  • disease: The disease is "agreed" among sources.
  • cell type: The cell type is "agreed" among sources.
  • Material type: The material is "agreed" among sources.
  • synonyms: This field is built using all the accessions and synonyms from all sources.
  • curated: This field is used to annotate if the cell line has been curated by the team, the classes are not curated, ai curated, manual curated.

Note: The database is a tab-delimited file that can be easily read and search using pandas or GitHub table rendering.

Requirements

  • Python 3.x
  • Libraries: pandas, spacy, click, owlready2, scikit-learn
  • Spacy model: en_core_web_md

Installation

Install the required Python packages using pip:

pip install pandas spacy click owlready2 scikit-learn
python -m spacy download en_core_web_md

Scripts

pycls cell-passports-database

Summary

The cell-passports-database command reads a CSV file containing cell passport data, filters it to include only cell lines, processes and renames specific columns, and then writes the processed data to an output file in tab-separated format.

Example Usage

pycls cell-passports-to-database --cell-passports path/to/cell_passports.csv --output path/to/output.tsv

Inputs

  • cell-passports: path to the folder containing the cell passport files
  • output: path to the output file

Flow

  • Read the CSV file specified by cell_passports.
  • Filter the data to include only rows where model_type is "Cell Line."
  • Select and rename specific columns.
  • Fill missing values with "no available".
  • Convert the age_at_sampling column to integers where applicable.
  • Write the processed data to the specified output file in tab-separated format.

Final Output

The final output is a TSV file containing the processed cell passport data cell-passports-db.tsv.

pycls ae-database

Summary

The ea-database command creates a database of cell lines from Expression Atlas files. It reads multiple TSV files from a specified folder, processes the data to remove duplicates, and aggregates information about each cell line. The function then checks this data against a provided cell line catalog from expression atlas, updates the database accordingly, and writes the final database to an output file in TSV format.

Example Usage

pycls ea-database --ea-folder path/to/ea_folder --ea-cl-catalog path/to/ea_cl_catalog.csv --output path/to/output.tsv

Inputs

  • ea_folder: Path to the folder containing Expression Atlas files.
  • ea_cl_catalog: Path to the Expression Atlas cell line catalog CSV file.
  • output: Path to the output file where the database will be saved.

Flow

  • Read all TSV files from the specified ea_folder.
  • Process each file to remove duplicates and aggregate cell line information.
  • Compare and update the aggregated data with the cell line catalog.
  • Write the final database to the specified output file in TSV format.

Final Output

The final output is a TSV file containing the processed cell line data from Expression Atlas.

pycls cellosaurus-database

Summary

The cellosaurus-database command creates a CelloSaurus database by parsing a gzipped CelloSaurus file and mapping its data to the BTO and Cell type ontologies. It filters the data based on specified species and writes the processed data to an output file.

Example Usage

pycls cellosaurus-database --cellosaurus path/to/cellosaurus.gz --output path/to/output.txt --bto path/to/bto.obo --cl path/to/cl.obo --filter-species "Homo sapiens,Mus musculus"

Inputs

  • cellosaurus: Path to the gzipped CelloSaurus database file.
  • output: Path to the output file where the processed database will be saved.
  • bto: Path to the BTO ontology file.
  • cl: Path to the Cell type ontology file.
  • filter-species: Optional, a comma-separated list of species to include in the output.

Flow

  • Read the BTO and Cell type ontology files using read_obo_file.
  • Parse the CelloSaurus file using parse_cellosaurus_file.
  • If filter_species is provided, filter the parsed data to include only the specified species.
  • Create new entries from the parsed CelloSaurus data using create_new_entry_from_cellosaurus.
  • Write the processed data to the output file using write_database_cellosaurus.

Final Output

The final output is a TSV of all filtered cell lines from Cellosaurus database.