About

SPLICEATOR v2.1: is a Donor and Acceptor Splice Sites prediction program based on Deep Learning approach. It used a Convolutional Neural Network with 3 layers trained in a wide range of species (>100), including both model and non-model organisms. Input sequences must be in fasta format with a minimum size of 10. Paste the sequences in the box or load your own .fasta file. For a high reliability have a threshold >95%.

DATA: The models are trained on a set of high quality sequences from the G3PO benchmark (Scalzitti et al., 2020). The dataset contains more than 20,000 sequences available in the 'Download' section.

IMPLEMENTATION: SPLICEATOR is implemented in python 3.7 and use Tensorflow 2.4.1 (Abadi et al., 2016) and Keras 2.3.1 (Chollet 2015).



Source code

You can found the source code following this link: Gitlab or in the 'Download' section.



How to cite us

Scalzitti, N., Kress, A., Orhand, R. et al. Spliceator: multi-species splice site prediction using convolutional neural networks. BMC Bioinformatics 22, 561 (2021)
https://doi.org/10.1186/s12859-021-04471-3


Statistics


Sources:
(Scalzitti et al., 2020) - A benchmark study of ab initio gene prediction methods in diverse eukaryotic organisms. BMC Genomics, 21, 293.
(Abadi et al., 2016) - TensorFlow: A system for large-scale machine learning. arXiv:1605.08695.
(Chollet 2015) - https://github.com/keras-team/keras