Consumer-focused section
- Date (published): **May 2, 2022
- Version (of this document): **v2.0
- Authors: **Diane Saunders, Heath Patterson, Liz McDonough, HuBMAP MxFBE assay team
- What is being measured? a. Tags*: Proteins; single-cell resolution; imaging b. Descriptive (optional): This spatial single-cell assay measures proteins. It is an Imaging-based spatial proteomics method. It provides quantitative and spatial analysis of over 40 protein markers across a whole tissue section at single-cell resolution.
- What analytical activities will the assay be used for within HuBMAP? a. Tags*: Map-Creation-for-Organs; Data Integration b. Descriptive (optional): […]
- What type of human samples are needed or used? a. Tags*: Fresh fozen, FFPE b. Link to sample metadata standards c. Descriptive (optional): […]
- Commercial Product: https://www.akoyabio.com/phenocycler/assays/** *Include tags that can be normalized across assays, allowing for assay filtering. When possible, use structured terminology (ontologies).
Assay Description
Primary Reference PMCIDs PMC8647621
Technology Overview
- CODEX is a strategy for generating highly multiplexed images using traditional fluorescent imaging. In brief, antibodies to antigens of interest are labeled with unique oligonucleotide barcodes. The barcoded antibodies are then applied to a tissue sample where they bind to target antigens. Using a microfluidics robot integrated with a fluorescent microscope, complementary oligonucleotide probes tagged with fluorophores are introduced to the tissue sample in groups of three antigen-specific probes at a time, allowing hybridization to the barcodes on the target antibodies. Tissue images are captured in each fluorescent channel, the fluorophore-tagged probes are gently removed, and then the process is repeated until all antigens have been visualized. DAPI is imaged during each cycle, enabling generation of a composite image with up to 50 protein targets co-registered on a single tissue section.
Key Definitions
- **Important terms used to describe CODEX data acquisition are defined and illustrated in the figures below.
Figure 1: Schematics illustrating CODEX-related terminology.
- **(A) The selected imaging region is automatically divided into tiles (individual fields of view).
- **(B) Images are acquired with adjacent tiles overlapping (e.g., 30%), as indicated by shaded regions, to enhance image alignment. Images are captured as the stage moves across the region row by row or via a serpentine path.
- **(C) Stitching is the process of aligning and merging tiles into a single composite image.
- **(D) Multiple z planes (depths) are acquired at each (x,y) tile position.
- (E) Segmentation algorithms, utilized during image processing, are used to visualize predicted cell boundaries based on nuclear and/or other cell morphology markers.
| Term | Definition |
|---|---|
| Alignment or registration | Process of transforming different images into one coordinate system (registration of all channels in each cycle is performed) |
| Autofluorescence | Endogenous fluorescence signal from tissue |
| Background subtraction | Subtraction of autofluorescence intensity from total intensity |
| Channels | Refers to the fluorescence excitation wavelengths used (e.g., 488, 540, 750); may also be by corresponding fluorophore names (e.g., DAPI, GFP, dsRED, Cy5) |
| CODEX | CO-Detection by inDEXing |
| Cycle | The process of adding three oligonucleotide-conjugated fluorophores to a tissue section in order to image the complementary-conjugated antibodies, then gently removing the fluorophores so a new set of antibodies can be visualized |
| Deconvolution | Process of reversing the optical distortion that takes place in an optical microscope to “sharpen” images and improve definition |
| Fluorescence | Light produced by a fluorophore that is bound to an oligonucleotide tag |
| Noise | Intensity not produced by light but electronic fluctuations or electronic background |
| Pitch | Distance between z-slices (images in z-stack) |
| Pixel | The smallest adjustable point of a rasterized image (resolution refers to the number of pixels in an image) |
| Region | User-defined imaging area |
| Signal | Intensity (detector counts) produced by fluorescence, both endogenous and introduced |
| Stitching | Process of combining multiple, overlapping fields of view (tiles) to generate a single, composite image of the tissue region |
| Tile | Rectangular field of view acquired at imaging magnification |
| X plane | Plane that determines width |
| Y plane | Plane that determines height |
| Z plane | Plane that determines depth |
| Z-stack | A series of images produced at different stage heights or positions along the z-axis |
Antibodies
- As relevant, include any details about antibody usage that are assay-specific.* Please see the HuBMAP standard report for antibody validation.
Provider-focused section
Directories and Files
Directory structure
- Structure the information as a table, exemplified below.
- When possible, an agreed-upon single assay-specific directory structure should be used rather than allowing for variable directory names with regular expressions (more conducive to downstream Data Consumer use).
- The directory structure should not include files. File definitions should happen in the “Files included” section where files can be more appropriately documented.
| Directory Name | Level | Required? | Description |
|---|---|---|---|
| raw/src | 1 | yes | This is the raw, unmodified files coming from the instrument (e.g., Akoya system). [Populated by the data provider.] |
| lab/drv | yes | Processed files produced by the lab that generated the data. [Populated by data providers.] | |
| lab/processed | 2 | yes | Image data that has been stitched and aligned and optionally has undergone background subtraction and deconvolution. |
| lab/segmented | 3 | no | Computationally predicted boundaries of cellular (nucleus, cytoplasm) and/or anatomical structures. |
| hive | yes | Processed files produced by HIVE using the common pipeline. [Populated by the HIVE.] | |
| hive/processed | 2 | yes | Image data that has been stitched and aligned and has undergone background subtraction and deconvolution. |
| hive/segmented | 3 | yes | Computationally predicted boundaries of cellular (nucleus, cytoplasm) and/or anatomical structures. |
Files Included
- Structure the information as a table, exemplified below.
- Files included (outside of the “lab” directory) should be agreed upon by the Assay Team and HIVE.
- *When possible, “file types” should include a link to an external definition, as exemplified below.
- When relevant, include a link to the program or pipeline used to generate each file. The program or pipeline used should be detailed in the “pipeline or data processing” metadata section below.
- If the program/pipeline will perform any QA/QC filtering of the data when generating the file, note this in the file description with additional details provided in the “Data processing pipeline” section below.
- Avoid regular expressions in file names unless absolutely necessary (e.g., to denote a batch of files as in a set of fastq files).
- Files containing the metadata should also be included when relevant, for example, the TSV with assay-level metadata, the antibodies TSV, a file with the pipeline parameters, etc.
| File | File type | Directory | Input file or precursor data | Generator program or pipeline with URL | Description |
|---|---|---|---|---|---|
| *.czi or *.scn | CZI or SCN | raw | n\/a | [Akoya link?] | Image data that is acquired directly from the microscope\; sometimes referred to as tiled or unstitched data. |
| *.tiff | TIFF | raw | n\/a | [Akoya link?] | Image data that is acquired directly from the microscope; sometimes referred to as tiled or unstitched data. |
| .OME.tiff | OME-TIF | lab \/ processed | n\/a | Images are a multi-page TIFF file comprised of all processed layers and metadata. | |
| HandE.tif or vHE.tif | TIFF | raw | n\/a | [Pathology Microscope] | H&E image or digital H&E image (if done). |
| antibodies .tsv | TSV | lab | n\/a | manual | Tab delimited file listing the set of all antibodies used. |
| lab-processing .tsv | TSV | lab | n\/a | manual | Comprehensive table containing the details of the lab-processing pipeline including all relevant parameters (if done). |
| dataset.json | JSON | lab | experiment .json (if generated) | manual | HuBMAP internal metadata standards that describe microscopy acquisition details for imaging spatial proteomics methods. |
Metadata
Sample-level
- Any assay-specific considerations for the sample-level metadata should be detailed here. This is a required documentation element. To avoid any confusion, you should explicitly state if there are no assay-specific considerations. Example fields that may warrant assay-level definitions are included below.
| Sample field name | Sample type [block; section; suspension] |
Definition |
|---|---|---|
| Processing time | ||
| Source storage time |
Assay-level
- This is the assay-specific metadata that’s included in the assay metadata TSV files.
- Please include full descriptions.
- Structure the information as a table, exemplified below.
- *THE FOLLOWING TABLE IS AN EXAMPLE — EDIT AS APPROPRIATE.
| Field | Required? | Data type | Description |
|---|---|---|---|
Assay-level categorical field values
- Categorical field options should be listed in the following table. *As the list will change over time, please coordinate the categorical lists with the HIVE.
| Field [from above] | Values [semicolon separated] |
|---|---|
Antibody
- Link to the Antibody TSV file.
HIVE data processing pipeline
- This section is to be completed by the HIVE.
- All pipeline processing steps should be detailed in the table below, including all parameter values used, as exemplified below.
- A figure should be included, as relevant, to better elucidate the pipeline levels, with each level being fully described in the table.
- Yes/No — The pipeline will produce the processed files from raw without human intervention. If “No”, then the required steps (human interventions) need to be detailed here.
- This processing should detail any expected pre-processing of input file(s)?
- The description should include what the processing step achieves.
- Any manual interventions should be documented with links to publications, as relevant.
| Level | Program | Version | Input File | Description |
|---|---|---|---|---|
| 1 | CIM software | 1.2 | Level 0 (.czi) | - What pre-processing of input file is expected? - What does this processing step achieve? - Are there any manual interventions before output is finalized? If yes, provide link to publication that has an explanation. |
Lab data processing pipeline
- The same details as provided in the above section (“HIVE data processing pipeline”) should be detailed for each lab that uploads data that is processed independently from the HIVE.
- Do not include the lab-processing details here; but rather include this lab-specific processing table of information with your data upload.
- In the Files section, describe any files that include the details of the lab-processing pipeline.