Update README with new dataset descriptions

README.md

@@ -5,23 +5,23 @@
 ---
 
 # Dataset Card for Dataset Name
-The `nucleotide_transformer_downstream_tasks` dataset features the 18 downstream tasks presented in the Nucleotide Transformer paper. They consist of both binary and multi-class classification tasks that aim at providing a consistent genomics benchmark.
+The `nucleotide_transformer_downstream_tasks` dataset features the 18 downstream tasks presented in the [Nucleotide Transformer paper](https://www.biorxiv.org/content/10.1101/2023.01.11.523679v3). They consist of both binary and multi-class classification tasks that aim at providing a consistent genomics benchmark.
+
+**We note that this is an updated version of this benchmark after the paper has been through peer-review. We highly encourage to move to this version in detriment of the [older version](https://huggingface.co/datasets/InstaDeepAI/nucleotide_transformer_downstream_tasks).**
 
 
 ## Dataset Description
 
 - **Repository:** [Nucleotide Transformer](https://github.com/instadeepai/nucleotide-transformer)
-- **Paper:** [The Nucleotide Transformer: Building and Evaluating Robust Foundation Models for Human Genomics](https://www.biorxiv.org/content/10.1101/2023.01.11.523679v1) 
+- **Paper:** [The Nucleotide Transformer: Building and Evaluating Robust Foundation Models for Human Genomics](https://www.biorxiv.org/content/10.1101/2023.01.11.523679v3) 
 
 ### Dataset Summary
 
-The different datasets are collected from 4 different genomics papers: 
-- [DeePromoter: Robust Promoter Predictor Using Deep Learning](https://www.frontiersin.org/articles/10.3389/fgene.2019.00286/full): The datasets features 3,065 TATA promoters and 26,532 non-TATA promoters, with each promoter yielding a negative sequence by randomly sampling parts of the sequence. The `promoter_all` dataset will feature all the promoters and their negative counterparts, while the `promoter_tata` and `promoter_no_tata` respectively provide the TATA and non-TATA parts of the dataset.
-- [A deep learning framework for enhancer prediction using word embedding and sequence generation](https://www.sciencedirect.com/science/article/abs/pii/S0301462222000643): To build the training dataset, the authors collect 742 strong
-enhancers, 742 weak enhancers and 1484 non-enhancers, and augment the dataset with 6000 synthetic enhancers and 6000 synthetic non-enhancers produced with a generative model. The test dataset is comprised of 100 strong enhancers, 100 weak enhancers and 200 non enhancers. The original paper uses this dataset to do both binary classification (i.e a sample gets classified as non-enhancer or enhancer) and 3-class classification (i.e a sample gets classified as non-enhancer, weak enhancer or strong enhancer). Both tasks are respectively tackled in the `enhancers` and `enhancers_types` datasets. 
-- [SpliceFinder: ab initio prediction of splice sites using convolutional neural network](https://pubmed.ncbi.nlm.nih.gov/31881982): The authors introduce a dataset containing 10,000 samples of donor site, acceptor site, and non-splice-site, resulting in 30,000 total samples that are featured in the `splice_sites_all` dataset.
-- [Spliceator: multi-species splice site prediction using convolutional neural networks](https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-021-04471-3): Two datasets are introduced by this paper, each of them contain splice sites and their corresponding negative datasets. The dataset `splice_sites_acceptor` features acceptor splice sites and the other, `splice_sites_donor`, donor splice sites.
-- [Qualitatively predicting acetylation and methylation areas in DNA sequences](https://pubmed.ncbi.nlm.nih.gov/16901084/): The paper introduces a set of datasets featuring epigenetic marks identified in the yeast genome, namely acetylation and metylation nucleosome occupancies. Nucleosome occupancy values in these ten datasets were obtained with Chip-Chip experiments and further processed into positive and negative observations to provide the datasets corresponding to the following histone marks:  `H3`, `H4`, `H3K9ac`, `H3K14ac`, `H4ac`, `H3K4me1`, `H3K4me2`, `H3K4me3`, `H3K36me3` and `H3K79me3`
+The different datasets are collected from the following resources: 
+- [ENCODE](https://www.encodeproject.org/): Histone ChIP-seq data for 10 histone marks in the K562 human cell line were obtained from ENCODE. We downloaded bed narrowPeak files with the following identifiers: H3K4me3 (ENCFF706WUF), H3K27ac (ENCFF544LXB), H3K27me3 (ENCFF323WOT), H3K4me1 (ENCFF135ZLM), H3K36me3 (ENCFF561OUZ), H3K9me3 (ENCFF963GZJ), H3K9ac (ENCFF891CHI), H3K4me2 (ENCFF749KLQ), H4K20me1 (ENCFF909RKY), H2AFZ (ENCFF213OTI). For each dataset, we selected 1kb genomic sequences containing peaks as positive examples and all 1kb sequences not overlapping peaks as negative examples.
+- [Screen](https://screen.wenglab.org/): Human enhancer elements were retrieved from ENCODE's SCREEN database. Distal and proximal enhancers were combined. Enhancers were split in tissue-specific and tissue-invariant based on the vocabulary from [Meuleman et al.](https://www.meuleman.org/research/dhsindex/). Enhancers overlapping regions classified as tissue-invariant were defined as that, while all other enhancers were defined as tissue-specific. We selected 400bp genomic sequences containing enhancers as positive examples and all 400bp sequences not overlapping enhancers as negative examples. We created a binary classification task for the presence of enhancer elements in the sequence (Enhancer) and a multi-label prediction task with labels being tissue-specific enhancer, tissue-invariant enhancer or none (Enhancer types).
+- [Eukaryotic Promoter Database](https://epd.expasy.org/epd/): We downloaded all human promoters from the Eukaryotic Promoter Database, spanning 49bp upstream and 10bp downstream of transcription start sites ([file](https://epd.expasy.org/ftp/epdnew/H_sapiens/006/Hs_EPDnew_006_hg38.bed)). This resulted in 29,598 promoter regions, 3,065 of which were TATA-box promoters (using the motif annotation at [https://epd.expasy.org/ftp/epdnew/H_sapiens/006/db/promoter_motifs.txt](https://epd.expasy.org/ftp/epdnew/H_sapiens/006/db/promoter_motifs.txt)). We selected 300bp genomic sequences containing promoters as positive examples and all 300bp sequences not overlapping promoters as negative examples. These positive and negative examples were used to create three different binary classification tasks: presence of any promoter element (Promoter all), a promoter with a TATA-box motif (Promoter TATA) or a promoter without a TATA-box motif (Promoter non-TATA).
+- [GENCODE](https://www.gencodegenes.org/human/release_44.html): We obtained all human annotated splice sites from GENCODE V44 gene annotation. Annotations were filtered to exclude level 3 transcripts (automated annotation), so all training data was annotated by a human.pancies. Nucleosome occupancy values in these ten datasets were obtained with Chip-Chip experiments and further processed into positive and negative observations to provide the datasets corresponding to the following histone marks:  `H3`, `H4`, `H3K9ac`, `H3K14ac`, `H4ac`, `H3K4me1`, `H3K4me2`, `H3K4me3`, `H3K36me3` and `H3K79me3`
 
 ## Dataset Structure
 ```