Dataset release
Hi all,
Happy New Year!
Just before the holidays, I had 2-3 people send me direct email asking about dataset release: where do I send it, how do I do it, etc. This is also something that came up in several of the Survey responses (sorry it's taking so long to put something together on that....trust me, I haven't forgotten!).
Here's my personal take:
My fundamental goal is that data gets released, period. So, I don't honestly care on a fundamental level where it gets released: the major places are still FlowRepository, and Immport. Mendeley gets a number of CyTOF papers, as do Figshare and Zenodo to a lesser extent.
My preference would *not* be a personal lab website, as it's too easy for those links to break (as we saw with a lot of the early CyTOF datasets initially hosted on Dana Pe'er's lab website that didn't get migrated). Remember, there are still print/PDF versions of your paper, so dataset links that are *not* hosted in a repository are more likely to have eventual link rot than those that do. Ideally, the dataset link will have a DOI, which helps referencing as well as helps prevent link rot.
When I release data, I still generally try to release it through FlowRepository. I think it's probably the best-known, and does force at least a bit of data description, without the huge amount of work that it takes to release to Immport. I will say, I hate the current FRv1.0 interface: dropdown menus that scroll for literally pages is *not* the way to go! The way I usually do it is fill out as much as I can in the top section (funding agency, PMID if possible, basic description of the dataset), then dump the FCS files into the bottom section.
In the middle section, you can toss in most file types, so that's where I put a PDF of my gating strategy, an Excel of my panel (including titer), Word or PDF protocol, and a README for anything else I think is important.
I honestly don't worry about the fact that most of my MiFlowCyt scores are in the 20-25% range: the stuff I put in the middle satisfies most of the remaining info without pulling my hair out fighting the clunky interface.
Here's an example: http://flowrepository.org/experiments/1363 ( FR-FCM-ZYAJ ; 35% MiFlowCyt)
I have *no* idea where they are on FlowRepositoryv2.0. It's been almost a year since the call for proposals closed.
Mike
Happy New Year!
Just before the holidays, I had 2-3 people send me direct email asking about dataset release: where do I send it, how do I do it, etc. This is also something that came up in several of the Survey responses (sorry it's taking so long to put something together on that....trust me, I haven't forgotten!).
Here's my personal take:
My fundamental goal is that data gets released, period. So, I don't honestly care on a fundamental level where it gets released: the major places are still FlowRepository, and Immport. Mendeley gets a number of CyTOF papers, as do Figshare and Zenodo to a lesser extent.
My preference would *not* be a personal lab website, as it's too easy for those links to break (as we saw with a lot of the early CyTOF datasets initially hosted on Dana Pe'er's lab website that didn't get migrated). Remember, there are still print/PDF versions of your paper, so dataset links that are *not* hosted in a repository are more likely to have eventual link rot than those that do. Ideally, the dataset link will have a DOI, which helps referencing as well as helps prevent link rot.
When I release data, I still generally try to release it through FlowRepository. I think it's probably the best-known, and does force at least a bit of data description, without the huge amount of work that it takes to release to Immport. I will say, I hate the current FRv1.0 interface: dropdown menus that scroll for literally pages is *not* the way to go! The way I usually do it is fill out as much as I can in the top section (funding agency, PMID if possible, basic description of the dataset), then dump the FCS files into the bottom section.
In the middle section, you can toss in most file types, so that's where I put a PDF of my gating strategy, an Excel of my panel (including titer), Word or PDF protocol, and a README for anything else I think is important.
I honestly don't worry about the fact that most of my MiFlowCyt scores are in the 20-25% range: the stuff I put in the middle satisfies most of the remaining info without pulling my hair out fighting the clunky interface.
Here's an example: http://flowrepository.org/experiments/1363 ( FR-FCM-ZYAJ ; 35% MiFlowCyt)
I have *no* idea where they are on FlowRepositoryv2.0. It's been almost a year since the call for proposals closed.
Mike