Search bioRxiv⌕ Search

Biology subjects

Mostavi, M.

Publications and source records attributed to Mostavi, M..

2 recordsLinked to original sources
bioRxiv2020

Machine Learning and Deep Learning challenges for building 2'O site prediction

2'-O-methylation (2'O) is one of the abundant post-transcriptional RNA modifications which can be found in all types of RNA. Detection and functional analysis of 2'O methylation have become challenging problems for biologists ever since its discovery. This paper addresses computational challenges for building Machine Learning and Deep Learning models for predicting 2'O sites. In particular, the impact of sequence length containing 2'O site, embedding method and the type of predictive model are each investigated separately. 30 different predictive models are built and each showed the impact of the mentioned parameters. The area under the precision-recall and receiving operating characteristics curves are utilized to test imbalanced case scenarios in the real world. By comparing the performance of these models, it is shown that embedding methods are crucial for Machine Learning models. However, they do not improve the performance of Deep Learning models. Furthermore, the best predictive model was further investigated to extract significant nucleotides surrounding 2'O sites. Interestingly, based on the significant score matrix achieved by all 2'O samples, it is depicted that model pays the highest attention at the location that the dominant 2'O motifs exist. Dataset and all of the codes are available at https://github.com/MMostavi/2_O_Me_sitePred

bioinformatics↗
bioRxiv2020

Predicting sites of epitranscriptome modifications using unsupervised representation learning based on generative adversarial networks

Epitranscriptome is an exciting area that studies different types of modifications in transcripts and the prediction of such modification sites from the transcript sequence is of significant interest. However, the scarcity of positive sites for most modifications imposes critical challenges for training robust algorithms. To circumvent this problem, we propose MR-GAN, a generative adversarial network (GAN) based model, which is trained in an unsupervised fashion on the entire pre-mRNA sequences to learn a low dimensional embedding of transcriptomic sequences. MR-GAN was then applied to extract embeddings of the sequences in a training dataset we created for eight epitranscriptome modifications, including m6A, m1A, m1G, m2G, m5C, m5U, 2'-O-Me, Pseudouridine ({Psi}) and Dihydrouridine (D), of which the positive samples are very limited. Prediction models were trained based on the embeddings extracted by MR-GAN. We compared the prediction performance with the one-hot encoding of the training sequences and SRAMP, a state-of-the-art m6A site prediction algorithm and demonstrated that the learned embeddings outperform one-hot encoding by a significant margin for up to 15% improvement. Using MR-GAN, we also investigated the sequence motifs for each modification type and uncovered known motifs as well as new motifs not possible with sequences directly. The results demonstrated that transcriptome features extracted using unsupervised learning could lead to high precision for predicting multiple types of epitranscriptome modifications, even when the data size is small and extremely imbalanced.

bioinformatics↗