Metadata Specification

MAPs

MAPs are tsne, umap, pca embeddings in 2D or 3D, or spatial maps for spatical data.

map is a list in the .json metadata file

For each map

"label" free text. Display label of the map, should be easily readable by users

"dataSubType" a string. Describe the nature of the map, must be embedding, spatial . Note this is the dataSubType attribute for the map, not the dataSubType attribute for the file

"dimension" a list of strings. They are the column headers of the dimension columns in the data file. They are used to retrieve data from db.

If it is a spatial map, there might be microscopy image(s) associated with each map.

"unit" (optional, only relevant to spatial map) a string. The unit of map values, e.g. pixel, micrometer

"micrometer_per_unit" (optional, only relevant to spatial map) a floating point number. The physical size in micrometer (µm) of value =1 in spatial map. The parameter will be used in rendering scale bar is the spatial map. If not specified, scale will not be shown.

"spot_diameter" (optional, relevant to spatial map) a floating point number of the size of spot in map unit (not image unit). The parameter will be used to determine sphere size shown in spatial map. If not specified, the size of the sphere will be determined by the browser.

"image" (optional, only relevant to a spatial map with associated images) an array of images, each image has its own parameters. See below.

For each image

"label" free text. Display label of the image, should be easily readable

"path" file path to the image file

"offset" an array of integers. Image offset in pixel in x and y dimension. See below for conversion from spatial coordinate values in map to pixel position in this image.

"image_scalef": floating point number. A scaling factor that converts spatial coordinate values in the spatial map (e.g. pixel or micrometer) to the pixel unit in this image. It works together with the "offset" parameter to convert spatial coordinate values in the spatial map to the actual pixel positions in this image.

• pixel_in_image_x = image_scalef * spatial_coordinate_x + offset_x

• pixel_in_image_y = image_scalef * spatial_coordinate_y + offset_y

"transcript" (optional, only relevant to a spatial transcriptomics map) an array of transcript data, each has its own json parameters. See below.

Note. The transcript coordinate must be in the same unit and scale as the map. Therefore no scaling or offset are needed to convert transcript coordinate and map coordinate.

"label" free text. Display label of the transcript data, should be easily readable

"path" file path to the transcript datafile

"dimension" a list of strings. Must have the same number of dimensions as the map. They are the column headers of the dimension columns in the transcript file. They are used to retrieve data from db.

Map examples

Example, map without image

Example, spatial map with matching microscopy image

Example, spatial map with matching microscopy image and transcript data

SURVIVAL TIME UNIT

Surival time unit is displayed on the x-axis of the KM plot. You specify it in metadata file under the "units" attribute.

"units" free text , KM plot x-axis unit, e.g. years, months, days

example

CUSTOM CATEGORICAL PHENOTYPE

You customize the display of features in a phenotype file ("type": "clinicalMatrix") by adding a "clinicalFeature" file and accompanying .json file. To do this there are two steps

1. Add a "clinicalFeature" reference to the .json file that accompanies the phenotype file. Note the colon notation in example below.

2. Compose the clinicalFeaure file (tab delimitated) and its .json file.

Below is an example for adding "clinicalFeature" reference in the phenotype file .json metadata

Below is an example for clinical feature file. This file is tab-delimitated, with headers "feature", "attribute", "value". The attributes are "valueType", "state", and "stateOrder". The values for the attribute "valueType" can be "category" or "float".

Below is an example clinicalFeature file .json file (clinicalFeature.txt.json)

An example of setting binary 0/1 variables as categorical data in the phenotype file. Xena automatically assumes 0/1 is numerical data when it is loaded, so if you want this data to be displayed as categorical you need to indicate it in the clinicalFeature file.

Below is an example of how you would do this for a feature called "your_featureName".

COLUMN DISPLAY NORMALIZATION

For genomic data matrix, the optional metadata parameter colNormalization sets the default display scale. If not specified, the browser automatically determines the scale.

colNormalization: ‘true’ | ‘log2(x)’ | ‘normal2’

• true: display centered by column mean, x - column average, example usage is gene expression matrix that already log transformed.

• log2(x): display in log2(x+1) scale, example usage is gene expression count matrix

• normal2: display value of 2 in the background color (i.e. white), typically used for copy number data where the normal = 2

example