neuralbioinfo/bacterial_promoters

Dataset Card for "Promoter Validation Data"

Overview

This dataset contains promoter sequences used for validation purposes in genetic research, focusing on prokaryotic promoters. It serves to facilitate the study of gene expression regulation, providing a comprehensive set of promoter sequences from various organisms.

The dataset comprises known promoter sequences from 75 organisms, retrieved from the Prokaryotic Promoter Database (PPD), alongside non-promoter sequences obtained from the NCBI RefSeq database (specifically sampled from CDS regions). It also includes non-promoter sequences constructed via higher and zero-order Markov chains that mirror compositional characteristics of known promoters. Additionally, an independent test set, focusing on E. coli sigma70 promoters, was employed, curated by Cassiano and Silva-Rocha (2020).

A balanced distribution approach was adopted to even out the number of positive and negative samples, with the dataset being systematically divided into training, validation, and test subsets. This stratification underpins a thorough evaluation of the model efficacy.

Figure 1: Promoter dataset - overview

The prokaryotic promoter sequences are typically 81bp long, ensuring compatibility with most tools' input prerequisites, particularly around the putative TSS region interval $[-60, +20]$. Our positive dataset encompasses promoter sequences from various species, predominantly found on both chromosomes and plasmids. Promoters included in the independent test set, based on exact match, were excluded from the training data. Species and contigs were mapped to NCBI assembly and sequence accessions.

To curate comprehensive non-promoter sequences (negative samples), we employed three strategies:

1. Using non-promoter sequences (CDS -- Coding Sequences).
2. Random sequences generated with a 3rd-order Markov chain.
3. Pure random sequences (0-order Markov chain) as proposed by Cassiano and Silva-Rocha (2020).

The distribution of this composite dataset was 40% CDS, 40% Markov-derived random sequences, and 20% pure random sequences (0-order Markov chain). One practical application of promoter detection in coding sequences is to check whether an unintentional promoter is injected or can be located inside a modified or designed coding sequence region, causing disruption. To cover this use-case, we incorporated the coding regions into our training and evaluation dataset. The CDS sequences were extracted from the genomic sequences of contigs, based on annotations from NCBI. The 81bp long CDS region samples were selected based on the NCBI-provided annotations for the available contigs with respect to the underlying species. The promoter regions often contain AT-rich sequences, i.e., TATA box. To capture and model the AT-rich regions, we applied 3rd and 0 order Markov chains to generate sequence examples that reflect the compositional property of known promoters.

Dataset Description

Data Collection and Processing

• Data Source: The positive samples, known promoters, are primarily drawn from the Prokaryotic Promoter Database (PPD), containing experimentally validated promoter sequences from 75 organisms. Non-promoter sequences are obtained from the NCBI RefSeq database, sampled specifically from CDS regions.
• Preprocessing: The dataset includes non-promoter sequences constructed via higher and zero-order Markov chains, which mirror compositional characteristics of known promoters. An independent test set based on E.coli sigma70 promoters is also included.

Dataset Structure

• Dataset Splits: The dataset is systematically divided into training, validation, and test subsets.
• Data Fields:
  • segment_id: Unique identifier for each segment.
  • ppd_original_SpeciesName: Original species name from the PPD.
  • Strand: The strand of the DNA sequence.
  • segment: The DNA sequence of the promoter region.
  • label: The label indicating whether the sequence is a promoter or non-promoter.
  • L: Length of the DNA sequence.
  • prom_class: The class of the promoter.
  • y: Binary label indicating the presence of a promoter.

Data Partitioning and Utilization

To ensure comprehensive evaluation, the dataset was split randomly into three parts:

1. Training set: 80% of the total data for model development and training.
2. Validation set: 10% of the data, aiding in fine-tuning model parameters.
3. Test set: Remaining 10% of the data, crucial for unbiased model performance evaluation.

Dataset Splits

• Training Set: Primary dataset used for model training.
• Test Set (Sigma70): Independent test set focusing on E.coli sigma70 promoters.
• Multispecies Set: Additional test set including various species, ensuring generalization across different organisms.

Use Cases & Limitations

This dataset is intended for use in training models for promoter prediction and understanding gene expression regulation. Limitations include the specific focus on prokaryotic promoters and the method of sequence generation using Markov chains.

How to Use

You can load the dataset using the Hugging Face datasets library:

from datasets import load_dataset
dataset = load_dataset("neuralbioinfo/bacterial_promoters")

Contact Information

For any questions, feedback, or contributions regarding the datasets or ProkBERT, please feel free to reach out:

• Name: Balázs Ligeti
• Email: obalasz@gmail.com

We welcome your input and collaboration to improve our resources and research.

Citation

@Article{ProkBERT2024,
  author  = {Ligeti, Balázs et al.},
  journal = {Frontiers in Microbiology},
  title   = {{ProkBERT} family: genomic language models},
  year    = {2024},
  volume  = {14},
  URL     = {https://www.frontiersin.org/articles/10.3389/fmicb.2023.1331233},
  DOI     = {10.3389/fmicb.2023.1331233}
}