Data Overview

The HTA is committed to ensuring access to spatial profiling data and is involved in multiple projects to establish standards and public access for multiplex data. Our efforts include creating and documenting our data analysis pipelines, developing metadata standards, making images available, and providing access to large primary datasets through Amazon Web Services (AWS) S3 buckets.

Below we describe our software pipeline, how data can be viewed online via MINERVA, and how to access primary full-resolution images.


Data Pipeline | Viewing Data Online with MINERVA | Public Access to Primary Data | Access Primary Image Data:

Click to enlarge.


This graphic depicts the flow of data into the Harvard Tissue Atlas (HTA). Data arise from collaborations between HTA members (green) and additional collaborators (yellow). Before depositing data into the atlases, the data goes through sequential databases, experimental methods, and interoperable software. Different roles within the workflow are numbered for clarity – note that a single person might hold multiple roles.

First, tissue samples from collaborators (Role 2) are onboarded into the LSP by a clinical research coordinator (Role 3) and data curator (Role 4). These de-identified samples are cataloged into internal LSP databases (Experiment Tracker and Reagent Tracker) and combined with a detailed plan for how the sample will be stained. The experimental team (Roles 5 and 6) then stains and images the whole slide, resulting in up to 1,000 high-magnification image tiles per slide.

A computational biologist (Role 7) extracts data from these image tiles with MCMICRO, a microscopy software pipeline developed by the LSP. The MCMICRO modules use algorithms to combine the tiles into a single high-plex mosaic image, segment the mosaic image (to isolate single cells), and extract image data (like single-cell intensity and cell morphology), which is recorded in a Spatial Cell Feature Table. Close collaboration with the project owner and pathologist (Roles 2 and 8) is required to assess the resulting data quality at various stages of analysis. Watch this video to learn more about MCMICRO.

Primary data is deposited onto online databases that allow the project team to dynamically view the data and collaborate on analyses. A data manager (Role 9) oversees access to these private data portals on Sage Synapse. At this point, project owners can add additional data types into the databases - consolidating data in one place facilitates subsequent collaboration and analysis. Watch this video to learn more about Synapse, OMERO, and related tools.

Now the scientific discovery process begins. New biological insights require both human inspection of the images and quantitative analysis of the single-cell data. This is highly collaborative (Roles 1, 2, 7, and 8) and is where the greatest innovation occurs! View two video examples of how multiplexed data come together for biological insights.

A goal of the HTA is to support data sharing. We release data to the public first through Minerva stories, which allow the project team to narrate a data-driven story and users to pan and zoom through the images manually without needing to download the data. Watch this video to learn more about Minerva. After publication, HTA data are deposited into existing public repositories or made available using public cloud storage systems.


MINERVA is a suite of software tools for visualizing, annotating and sharing large, multi-dimensional images in a browser using digital docents and narration without the need to download the data. MINERVA enables interactive viewing and efficient data annotation. It can also display other types of data visualization. MINERVA does not require the installation of any software and browsing is anonymous. Nearly all HTA data are available as one of two types of MINERVA-based visualization, described below.


Data Explorations are MINERVA stories that guide users through the complexities of a large dataset and analysis by making extensive use of narration and waypoints. We aim to develop at least one exploration per HTA publication or Atlas dataset.

Data image
Pathological annotations - Multiplexed 3D atlas of state transitions and immune interactions in colorectal cancer - CRC01
Data image
Graphical Abstract - Multiplexed 3D atlas of state transitions and immune interactions in colorectal cancer
Data image
Multiplexed 3D atlas of state transitions and immune interactions in colorectal cancer (CRC01 Z-stack)


Data Overviews are MINERVA stories for a single, minimally-processed (Level 2 or 3) whole slide image with no annotation or analysis. Data Overviews provide online viewing access to largely unprocessed image data; MCMICRO will soon generate data overviews automatically as a component of the pipeline.


A key goal of the HTA is to provide access to primary data consistent with FAIR Principles. This remains a work in progress as we develop and implement repositories and standards for our diverse data types. As a result, not all data sets are currently available for download at this time.

All NCI Human Tumor Atlas Network data are released pre-publication via its data portal. We are still working to release datasets associated with other funding mechanisms via public repositories.


Images and metadata will be available in an AWS S3 bucket at a location specified for each paper’s data availability statement. For up-to-date instructions on how to download image data found on this website, please visit our access instructions on Zenodo (DOI: 10.5281/zenodo.10223574). Email tissue-atlas(at) if you have questions or experience issues.