Dear @scRNAseqandscATACseqDataAnalysisDREAMChallengeOrganizers,
so I registerd for the challange, but I am only able to see the example data of PBMC3K cells.
Where do I find the data for the challange?
Thank you in advance.
Created by Zoe ZoeJ Hi @bahrad,
> would we provide a pretrained model, i.e., load saved model weights, based on these other public datasets, which is then evaluated against DS.1.C in the leaderboard phase and other datasets in some other later phase?
This is correct. You would submit the pre-trained model that will be evaluated using DS.1.C for leaderboard phase. Sorry for the confusion.
Thank you for your interest. We are a little confused by the wording in the Data Description that says: "The following is the example of directory structure containing the training data files which will be available as CSV files to use in Docker:". Does this mean that we are supposed to train a model within Docker on that data set (DS.1.C)? Or would we provide a pretrained model, i.e., load saved model weights, based on these other public datasets, which is then *evaluated* against DS.1.C in the leaderboard phase and other datasets in some other later phase? Thanks! Hi @svenrahmann,
Thank you for your interest in the scRNA-seq and scATAC-seq Data Analysis DREAM Challenge!
> I also thought that you were going to provide the other datasets
This challenge does NOT provide the real datasets described in [Data wiki page](https://www.synapse.org/#!Synapse:syn26720920/wiki/620124) for the purpose of model building. For Task 1, participants could download _Seurat 3k PMBC_ count matrix [here](https://www.synapse.org/#!Synapse:syn48025824) or find other publicly available datasets to build the models.
> My understanding is that you have generated a lot of datasets for the challenge, but only a few of them are used in phase 1 or phase 2
The submitted models will be validated and evaluated with the real datasets on our server. In the current phase (leaderboard phase), a subset of the real datasets will be used to test the functionality of your algorithms. A larger/full portion of real datasets will be used to evaluate your model performance in final phase. ?Data preparation for challenge testing and validation? sections of the [Task1: scRNA-seq](https://www.synapse.org/#!Synapse:syn26720920/wiki/620137) and [Task 2: scATAC-seq](https://www.synapse.org/#!Synapse:syn26720920/wiki/620138) sub-wiki pages have some details on this.
Thank you! Hi,
now I'm confused. My understanding is that you have generated a lot of datasets for the challenge, but only a few of them are used in phase 1 or phase 2. I also thought that you were going to provide the other datasets. I am much less interested in the challenge if I have to go looking all over the internet for suitable example data.
Best regards!
Hi @ZoeJ,
Thank you for your interest in the scRNA-seq and scATAC-seq Data Analysis DREAM Challenge!
For this challenge, the training data will NOT be provided for model building. Instead please use publicly available datasets and/or those available to your institution(s) for this purpose, such as [Seurat 3k PMBC](https://www.synapse.org/#!Synapse:syn48025824) and [Multimodal Single-Cell Data Integration - the NeuralPS challenge](https://openproblems.bio/neurips_2021/) for Task 1.
Thank you!