Submitting data to the UCSC Cell Browser¶
At this time, we are happy to host pretty much any single-cell dataset, regardless of the library prepartion (10x, Smart-seq2, etc), organism (human, mouse, zebrafish, etc), or analysis method (Seurat, Scanpy, Monocle, etc).
A cell browser requires at minimum three things:
Expression matrix
Metadata with cell names and cluster field
2D Layout coordinates
If you can provide these alongside some description of the dataset, we’ll host it. Contact us at cells@ucsc.edu to get started.
Preparing and sharing your files¶
Before we can make a cell browser for you, you have to share the data with us. We accept the following file types:
Seurat RDS, Rdata, or Robj files
Scanpy h5ad or Loom files
A collection of tsv or csv files
The output directory of one of our cbImport* tools
After you have your data in one of the formats above, you will have to share the data with us.
We prefer data to be shared in a way that is easy for us to download with something like wget, so the following methods are ideal:
University or other institutional server space
AWS
GEO
Dropbox (if shared via link, e.g. https://www.dropbox.com/s/NN/my_seurat.rds)
We do accept files via other methods, although they take a little more work for us to move to our server. These include methods such as:
Box
Google Drive
Dropbox, other methods of sharing data
Other things we want from you¶
Dataset description¶
Alongside your submission, it would be great if you filled out a
desc.conf file. At
the very least, it should have the abstract, methods, and title filled out.
However, you are welcome to fill out more fields and make it as complete as
you would like. You can run cbBuild --init to copy an example desc.conf
into your current directory or you can copy one from
our Github repo.
If you have a collection of datasets, please provide basic desc.conf for each of the datasets in the collection. These don’t have to be as comprehensive as the top-level one for the collection itself. At minimum, they should include two tags: (1) title and (2) abstract. However, feel free to fill out as many of the desc.conf fields as you think is helpful for that dataset.
Dataset shortname¶
It would be great if you could suggest a dataset shortname at the time of your submission, although we’re happy to make one up for you. An ideal short name fulfills the following requirements
All lowercase
Words separated by dashes (“-“)
Four words or less (don’t be afraid to abreviate words, e.g. development -> dev)
Informative
A great example is cortex-dev - it’s all lowercase, the two words are separated by dashes, it’s short at only two words long, and informs you that the dataset is focused on cortex development. It fulfills all four points above.
Other great examples:
mouse-nervous-system
skeletal-muscle
mouse-oligo-het
covid-hypertension
The short name doesn’t have to be perfect, but good enough to communicate something about your dataset in a few words.
“Quick Genes”¶
This is a list of 10-15 genes that you think are important to your dataset(s). In addition to the list of gene symbols, it is great to have a word or two about why it’s in the list (e.g. “Fst, Paraxial Mesoderm”; “HES1, Fig1D”). If you have a collection of datasets, you can have one set of genes for every dataset in the collection or a different set for each.
This list and descriptions should be in CSV or TSV format.
Getting your URL¶
After you submit your dataset to us, we will import the data and make a preliminary version available on our development server. We will work with you to iterate and make improvements to this version first. Once you give your final approval, we will push the data to our main site, cells.ucsc.edu. Once there, you will recieve the final URL, e.g. cortex-dev.cells.ucsc.edu. This is the URL you should place in your paper, link to from your lab website, tweet about, etc. Please do not put the url to our development server in your paper, since it is under active development, we occasionaly break it.
FAQs¶
Can I keep my dataset private until a later date, but still accessible to reviewers?¶
Yes, we offer limited methods for keeping datasets private. We can hide datasets from being listed alongside the others we host. This means that someone would need to know the URL or dataset name to be able to access your dataset. For example, this means that someone would need the URL cells.ucsc.edu/?ds=cortex-dev or know the name (cortex-dev) to access the dataset.