sequence-to-function modeling

Sensitive alignment using paralogous sequence variants improves long read mapping and variant calling in segmental duplications

DuploMap improves long-read mapping and variant calling in segmental duplications by using paralogous sequence variants to distinguish alignments.

De Novo Protein Design for Novel Folds using Guided Conditional Wasserstein Generative Adversarial Networks (gcWGAN)

This paper develops a deep generative model, gcWGAN, to design novel protein folds using sequence-structure data, achieving higher fold coverage and sequence diversity than a competing method.

Developing a novel recurrent neural network architecture with fewer parameters and good learning performance

This paper develops a novel RNN architecture, YamRNN, with fewer parameters and faster learning performance compared to LSTM and GRU on sequence classification tasks.

DeepGraphMol, a multi-objective, computational strategy for generating molecules with desirable properties: a graph convolution and reinforcement learning approach

This paper introduces DeepGraphMol, a method using graph convolution and reinforcement learning to generate molecules with desired properties, such as binding to dopamine transporters while avoiding norepinephrine binding.

Ubiquitous Forbidden Order in R-group classified protein sequence of SARS-CoV-2 and other viruses

The paper identifies unique forbidden order patterns in viral protein sequences, suggesting a novel method for species characterization.

NanoReviser: An Error-correction Tool for Nanopore Sequencing Based on a Deep Learning Algorithm

NanoReviser improves nanopore sequencing accuracy using deep learning, reducing error rates by over 5% in E. coli and human datasets.

Site-specific amino acid distributions follow a universal shape

The paper shows that a single-parameter model can accurately predict amino acid frequency ranks across protein sequences, revealing a universal distribution shape.

A structured observation distribution for generative biological sequence prediction and forecasting

This paper introduces a structured observation distribution (MuE) to improve generative probabilistic modeling of biological sequences, showing it outperforms methods using multiple sequence alignment preprocessing in sequence prediction and forecasting.

Large-scale DNA-based phenotypic recording and deep learning enable highly accurate sequence-function mapping

The paper introduces DNA-based phenotypic recording to generate large-scale sequence-function datasets and uses deep learning to predict gene regulatory element function with high accuracy.

Consistent Consideration of RNA Structural Alignments Improves Prediction Accuracy of RNA Secondary Structures

This paper introduces ConsHomfold and ConsAlifold, which improve RNA secondary structure prediction by consistently considering global pairwise structural alignments, achieving better accuracy with comparable speed to conventional methods.

Automated design and optimization of multitarget schizophrenia drug candidates by deep learning

This paper develops a deep learning system to design multitarget antipsychotic drugs, generating novel compounds with activity against multiple receptors and demonstrating efficacy in animal models.

Predicting Antibody Developability from Sequence using Machine Learning

This paper develops a machine learning pipeline to predict antibody developability from sequence, achieving high sensitivity and specificity on a dataset of 2400 antibodies.

Transforming the Language of Life: Transformer Neural Networks for Protein Prediction Tasks

A RoBERTa-style Transformer, pre-trained on 450K Swiss-Prot sequences in 18 hours on 4 V100 GPUs, rivals DeepFam at family classification and beats PIPR at human protein-protein interaction prediction.

ComPotts: Optimal alignment of coevolutionary models for protein sequences

ComPotts aligns Potts models of proteins to improve pairwise sequence alignment by considering coevolutionary relationships.

Bio-informed Protein Sequence Generation for Multi-class Virus Mutation Prediction

This paper introduces a GAN-based model to generate protein sequences for predicting virus mutations across multiple classes.

SARS-CoV-2 sequence typing, evolution and signatures of selection using CoVa, a Python-based command-line utility

The paper introduces CoVa, a Python-based tool for analyzing SARS-CoV-2 genomes, identifying selection signatures, and defining sequence typing, with findings on geographic variation and mutation rates.

RNA secondary structure packages evaluated and improved by high-throughput experiments

The paper evaluates RNA structure prediction packages using synthetic data and introduces an improved model that performs better on diverse datasets.

Meta-Align: A Novel HMM-based Algorithm for Pairwise Alignment of Error-Prone Sequencing Reads

Meta-Align improves DNA sequence alignment by incorporating quality scores and handling frameshift errors using an HMM-based approach, outperforming TBLASTX in simulated benchmarks.

Restricted sequence variation in Streptococcus pyogenes penicillin binding proteins

The paper investigates PBP sequence variation in S. pyogenes and finds that mutations in PBPs are rare, with few amino acid differences across most isolates, contrasting with more frequent mutations in S. pneumococcus.

Leverging Deep Learning to Simulate Coronavirus Spike proteins has the potential to predict future Zoonotic sequences

This study uses deep learning to simulate coronavirus spike protein sequences, which show BLAST matches to known sequences and may aid in predicting future zoonotic variants.

Minimum epistasis interpolation for sequence-function relationships

The paper introduces a method to impute sequence-function relationships by minimizing epistasis, showing it provides a less epistatic fit with comparable predictive power.

Designing Feature-Controlled Humanoid Antibody Discovery Libraries Using Generative Adversarial Networks

This paper uses a GAN to generate diverse human antibody libraries with controlled properties, validated through phage display and biophysical evaluation.

Entropy-Based Analysis of Vertebrate Sperm Protamine Sequences: Evidence of Dityrosine and Cysteine-Tyrosine Cross-Linking in Sperm Protamines

The paper analyzes vertebrate sperm protamine sequences to identify conserved residues and proposes dityrosine or cysteine-tyrosine cross-linking as mechanisms for protamine structure in metatherians.

Deep Generative Models for 3D Compound Design

This paper introduces DeLinker, a 3D graph-based generative model for compound design that outperforms baselines by incorporating structural information.

Comparative analysis of novel MGISEQ-2000 sequencing platform vs Illumina HiSeq 2500 for whole-genome sequencing

This study compares the MGISEQ-2000 and Illumina HiSeq 2500 platforms for whole-genome sequencing, finding comparable error rates and performance in SNP detection.

DeCoDe: degenerate codon design for complete protein-coding DNA libraries

DeCoDe optimizes degenerate codon libraries for protein-coding DNA, outperforming existing methods and supporting mixed-length libraries.

Learning supervised embeddings for large scale sequence comparisons

This paper introduces SuperVec and H-SuperVec, supervised methods for learning sequence embeddings that improve retrieval performance and speed compared to existing approaches.

Large X-linked palindromes undergo arm-to-arm gene conversion across Mus lineages

The paper compares X-palindromic regions across Mus species and finds evidence of arm-to-arm gene conversion at rates similar to autosomal gene conversion, suggesting it drives both sequence homogenization and diversification.

A compact vocabulary of paratope-epitope interactions enables predictability of antibody-antigen binding

The paper identifies a compact vocabulary of paratope-epitope interaction motifs that enables machine learning prediction of antibody-antigen binding.

A method for predicting evolved fold switchers exclusively from their sequences

The paper identifies evolved fold switchers from amino acid sequences by analyzing inconsistencies in secondary structure predictions between helix and strand conformations.