Human Protein Atlas Subcellular Section

Dataset Overview

Data Type

Fluorescence microscopy images

Citation

Publication: A subcellular map of the human proteome. Science (2017) https://doi.org/10.1126/science.aal3321

Dataset: data available from https://v23.proteinatlas.org/

Dataset Card Authors

Chan Zuckerberg Initiative

Dataset Card Contact

Uses

Primary Use Cases

Investigate cellular localization of proteins in single cells

Out-of-Scope or Unauthorized Use Cases

Do not use the dataset for the following purposes:

• Usage not covered by the license.
• Any use that is not in accordance with the Acceptable Use Policy

Dataset Structure

Each protein was imaged using confocal microscopy in up to three cell lines, and occasionally there were multiple antibodies for one protein and each antibody was imaged once. For each condition, two or six fields-of-view were captured. In the JPEG version hosted on https://v23.proteinatlas.org/, each image is a field of view with four colors: red for microtubule, yellow for endoplasmic reticulum, blue for DNA and green for protein.

Personal and Sensitive Information

No personal and sensitive information is included.

Dataset Creation

Curation Rationale

This dataset was collected to visualize protein localization in single cells and capture their variation across single cells and cell lines.

Data Collection and Processing

The cells were fixed and stained with antibodies targeting the protein of interest, gamma tubulin (to visualize microtubules), calreticulin (to visualize the endoplasmic reticulum) and DAPI (to visualize the nuclei). All images were acquired with a confocal microscope in a single facility. Two or six (if the localization is highly variable from cell to cell) field-of-view images were taken for each protein in each cell line per antibody (if multiple), and two representative images were selected for display on the HPA website. For more details see: https://v23.proteinatlas.org/humanproteome/subcellular/method

Annotation process

Protein localization and corresponding confidence were manually annotated by trained experts. Protein localizations were categorized into 35 different organelles and subcellular structures based on their positions relative to other cellular structures imaged. Confidence for the localization annotation was assessed by taking into account the HPA data, how well different antibodies for the same protein compare, and how external data for the same protein agree with HPA data. More details see https://v23.proteinatlas.org/about/assays+annotation#if

Who are the annotators?

The annotation was done by the team who generated and processed the data.

Bias, Risks, and Limitations

Depending on specific use cases, be mindful of the following properties about this dataset:

• Not every protein is imaged in every cell line.
• The number of images varies among proteins and cell lines. Some proteins or cell lines have many more images than others.
• Some proteins have more than one antibody.
• Some antibodies target multiple proteins so the images from these antibody stainings are listed under each of its protein targets.
• The total amount of cells for each condition may be limited for certain types of statistical analysis, especially if cell-to-cell variability is large.
• The HPA website only hosts the JPEG version of images with a small compression due to conversion from TIFF to JPEG. The original TIFF images may be available upon request to contact@proteinatlas.org.

Acknowledgements

See source reference: https://www.science.org/doi/10.1126/science.aal3321