Machine Learning in Bioinformatics

DNABERT: pre-trained Bidirectional Encoder Representations from Transformers model for DNA-language in genome

DNABERT is a pre-trained transformer model for DNA sequences that achieves state-of-the-art performance on sequence prediction tasks and enables visualization of nucleotide-level importance across species.

Nucleic Transformer: Deep Learning on Nucleic Acids with Self-attention and Convolutions

The paper introduces the Nucleic Transformer, a deep learning model for DNA/RNA tasks that is effective, interpretable, and performs well with limited data in promoter classification, viral genome identification, and mRNA vaccine degradation analysis.

Generating novel protein sequences using Gibbs sampling of masked language models

The paper generates novel protein sequences using Gibbs sampling of BERT-style masked language models and evaluates their quality against natural sequences.

dom2vec: Unsupervised protein domain embeddings capture domains structure and function providing data-driven insights into collocations in domain architectures

The paper introduces dom2vec, a method for learning protein domain embeddings and evaluates their quality using intrinsic strategies, showing they perform well on downstream tasks.

Embeddings from deep learning transfer GO annotations beyond homology

This paper proposes using deep learning embeddings to transfer GO annotations based on protein sequence proximity rather than homology, achieving performance comparable to existing methods.

Capsule network for protein ubiquitination site prediction

The paper introduces a deep learning model called Caps-Ubi for predicting ubiquitination sites across species, achieving high accuracy and outperforming existing predictors.

Prediction of GPI-Anchored proteins with pointer neural networks

The paper introduces NetGPI, a new method for predicting GPI-anchored proteins using recurrent neural networks with attention mechanisms, achieving better performance than existing methods.

The power of universal contextualised protein embeddings in cross-species protein function prediction

The paper demonstrates that using universal contextualized protein embeddings from SeqVec allows effective cross-species molecular function prediction, especially for understudied species.

A Sequence-based Antibody Paratope Prediction Model Through Combing Local-Global Information and Partner Features

This paper proposes a sequence-based method for predicting antibody paratopes by integrating local-global sequence information and partner antigen features, achieving better performance than existing methods.

Representation learning applications in biological sequence analysis

This paper reviews the application of representation learning in biological sequence analysis, focusing on how embedding techniques transform sequences for functional and structural insights.

Light Attention Predicts Protein Location from the Language of Life

The paper introduces a lightweight deep learning method using protein language model embeddings to predict protein subcellular localization, outperforming state-of-the-art methods without using multiple sequence alignments.

Getting to know each other: PPIMem, a novel approach for predicting transmembrane protein-protein complexes

PPIMem predicts transmembrane protein-protein complexes by identifying interaction motifs from 3D structures and searching them in databases, predicting 21,544 complexes across 39 species.

Neural network and random forest models in protein function prediction

The paper develops an ensemble system combining random forest and neural network models to predict protein functions using sequence and taxonomic features, showing competitive performance in CAFA3 evaluations.

A convNet based multi-label microRNA subcellular location predictor, by incorporating k-mer positional encoding

The paper introduces a kmerRP2vec feature representation method combined with CNN and RNN to predict miRNA subcellular locations, outperforming existing methods on a public dataset.

DeepCSO: a deep-learning network approach to predicting Cysteine S-sulphenylation sites

This study develops DeepCSO, a deep-learning tool for predicting Cysteine S-sulphenylation sites across species, showing improved performance over existing methods.

MpsLDA-ProSVM: predicting multi-label protein subcellular localization by wMLDAe dimensionality reduction and ProSVM classifier

The paper introduces MpsLDA-ProSVM, a model for predicting multi-label protein subcellular localization using feature refinement and a novel classifier, achieving high accuracy across multiple organisms.

Protein Cargo Assessment through Residue Alterations

The paper introduces TransSite, a computational method to assess the impact of single residue changes on protein targeting sequences, identifying recurring mutations in phenotype-related proteins across human and non-human data.

StackPDB: predicting DNA-binding proteins based on XGB-RFE feature optimization and stacked ensemble classifier

StackPDB predicts DNA-binding proteins using XGB-RFE feature optimization and a stacked ensemble classifier, achieving high accuracy and MCC on benchmark datasets.

CAMP: a Convolutional Attention-based Neural Network for Multifaceted Peptide-protein Interaction Prediction

CAMP predicts peptide-protein interactions and binding residues using a convolutional attention-based neural network, outperforming existing methods on benchmark datasets.

DHS-Crystallize: Deep-Hybrid-Sequence based method for predicting protein Crystallization

DHS-Crystallize predicts protein crystallization using a hybrid sequence-based method combining deep learning and physicochemical features, outperforming existing approaches.

SeqPIP-2020: Sequence based Protein InteractionPrediction Contest

The paper summarizes results from a competition to develop PPI predictive models using sequence data with high-quality datasets.

Novel transformer networks for improved sequence labeling in genomics

The paper introduces transformer networks for sequence labeling in genomics, achieving state-of-the-art performance on tasks like transcription start site annotation in E. coli.

Evaluation of Methods for Protein Representation Learning: A Quantitative Analysis

This study evaluates protein representation learning methods across three tasks: semantic similarity, function prediction, and drug target classification, highlighting their strengths and limitations.

CATH functional families predict protein functional sites

The paper introduces FunSite, a machine learning method for predicting functional sites in proteins using sequence, structure, and evolutionary data from CATH functional families, showing superior performance over existing methods.

A generalized machine-learning aided method for targeted identification of industrial enzymes from metagenome: a xylanase temperature dependence case study

This paper develops a machine learning method to predict the thermal dependence of xylanases from metagenomic data, identifying three novel xylanases from rumen microbiota.

TNFPred: Identifying tumor necrosis factors using hybrid features based on word embeddings

This paper introduces TNFPred, a model that identifies tumor necrosis factors from cytokine sequences using hybrid features based on word embeddings, achieving high accuracy in classification.

Prediction of Protein-Protein Interactions Based on L1-Regularized Logistic Regression and Gradient Tree Boosting

This paper introduces a PPI prediction method using gradient tree boosting and L1-regularized logistic regression, achieving high accuracy on multiple organisms.

T4SE-XGB: interpretable sequence-based prediction of type IV secreted effectors using eXtreme gradient boosting algorithm

This study introduces T4SE-XGB, an XGBoost-based model for predicting type IV secreted effectors from protein sequences, achieving high accuracy and interpretability.

DeepKhib: a deep-learning framework for lysine 2-hydroxyisobutyrylation sites prediction

The paper introduces DeepKhib, a deep-learning framework for predicting lysine 2-hydroxyisobutyrylation sites, showing improved performance over existing methods across multiple species.

FastSK: Fast Sequence Analysis with Gapped String Kernels

FastSK improves the speed and scalability of gapped k-mer kernel computations for sequence analysis tasks, outperforming existing methods in both speed and accuracy.