More Like this

1. Scalable preprocessing for sparse scRNA-seq data exploiting prior knowledge

   https://doi.org/10.1093/bioinformatics/bty293  

   Mukherjee, Sumit ; Zhang, Yue ; Fan, Joshua ; Seelig, Georg ; Kannan, Sreeram ( June 2018 , Bioinformatics)

   Single cell RNA-seq (scRNA-seq) data contains a wealth of information which has to be inferred computationally from the observed sequencing reads. As the ability to sequence more cells improves rapidly, existing computational tools suffer from three problems. (i) The decreased reads-per-cell implies a highly sparse sample of the true cellular transcriptome. (ii) Many tools simply cannot handle the size of the resulting datasets. (iii) Prior biological knowledge such as bulk RNA-seq information of certain cell types or qualitative marker information is not taken into account. Here we present UNCURL, a preprocessing framework based on non-negative matrix factorization for scRNA-seq data, that is able to handle varying sampling distributions, scales to very large cell numbers and can incorporate prior knowledge.

   We find that preprocessing using UNCURL consistently improves performance of commonly used scRNA-seq tools for clustering, visualization and lineage estimation, both in the absence and presence of prior knowledge. Finally we demonstrate that UNCURL is extremely scalable and parallelizable, and runs faster than other methods on a scRNA-seq dataset containing 1.3 million cells.

   Source code is available at https://github.com/yjzhang/uncurl_python.

   Supplementary data are available at Bioinformatics online.

    

   more » « less
2. Estimation of speciation times under the multispecies coalescent

   https://doi.org/10.1093/bioinformatics/btac679  

   Peng, Jing ; Swofford, David L. ; Kubatko, Laura ; Martelli, ed., Pier Luigi ( October 2022 , Bioinformatics)

   The multispecies coalescent model is now widely accepted as an effective model for incorporating variation in the evolutionary histories of individual genes into methods for phylogenetic inference from genome-scale data. However, because model-based analysis under the coalescent can be computationally expensive for large datasets, a variety of inferential frameworks and corresponding algorithms have been proposed for estimation of species-level phylogenies and associated parameters, including speciation times and effective population sizes.

   We consider the problem of estimating the timing of speciation events along a phylogeny in a coalescent framework. We propose a maximum a posteriori estimator based on composite likelihood (MAPCL) for inferring these speciation times under a model of DNA sequence evolution for which exact site-pattern probabilities can be computed under the assumption of a constant θ throughout the species tree. We demonstrate that the MAPCL estimates are statistically consistent and asymptotically normally distributed, and we show how this result can be used to estimate their asymptotic variance. We also provide a more computationally efficient estimator of the asymptotic variance based on the non-parametric bootstrap. We evaluate the performance of our method using simulation and by application to an empirical dataset for gibbons.

   The method has been implemented in the PAUP* program, freely available at https://paup.phylosolutions.com for Macintosh, Windows and Linux operating systems.

   Supplementary data are available at Bioinformatics online.

    

   more » « less
3. Identifying antimicrobial peptides using word embedding with deep recurrent neural networks

   https://doi.org/10.1093/bioinformatics/bty937  

   Hamid, Md-Nafiz ; Friedberg, Iddo ; Hancock, ed., John ( November 2018 , Bioinformatics)

   Antibiotic resistance constitutes a major public health crisis, and finding new sources of antimicrobial drugs is crucial to solving it. Bacteriocins, which are bacterially produced antimicrobial peptide products, are candidates for broadening the available choices of antimicrobials. However, the discovery of new bacteriocins by genomic mining is hampered by their sequences’ low complexity and high variance, which frustrates sequence similarity-based searches.

   Here we use word embeddings of protein sequences to represent bacteriocins, and apply a word embedding method that accounts for amino acid order in protein sequences, to predict novel bacteriocins from protein sequences without using sequence similarity. Our method predicts, with a high probability, six yet unknown putative bacteriocins in Lactobacillus. Generalized, the representation of sequences with word embeddings preserving sequence order information can be applied to peptide and protein classification problems for which sequence similarity cannot be used.

   Data and source code for this project are freely available at: https://github.com/nafizh/NeuBI.

   Supplementary data are available at Bioinformatics online.

    

   more » « less
4. DaTeR: error-correcting phylogenetic chronograms using relative time constraints

   https://doi.org/10.1093/bioinformatics/btad084  

   Mondal, Abhijit ; Rangel, L. Thiberio ; Payette, Jack G. ; Fournier, Gregory P. ; Bansal, Mukul S. ; Schwartz, ed., Russell ( February 2023 , Bioinformatics)

   A chronogram is a dated phylogenetic tree whose branch lengths have been scaled to represent time. Such chronograms are computed based on available date estimates (e.g. from dated fossils), which provide absolute time constraints for one or more nodes of an input undated phylogeny, coupled with an appropriate underlying model for evolutionary rates variation along the branches of the phylogeny. However, traditional methods for phylogenetic dating cannot take into account relative time constraints, such as those provided by inferred horizontal transfer events. In many cases, chronograms computed using only absolute time constraints are inconsistent with known relative time constraints.

   In this work, we introduce a new approach, Dating Trees using Relative constraints (DaTeR), for phylogenetic dating that can take into account both absolute and relative time constraints. The key idea is to use existing Bayesian approaches for phylogenetic dating to sample posterior chronograms satisfying desired absolute time constraints, minimally adjust or ‘error-correct’ these sampled chronograms to satisfy all given relative time constraints, and aggregate across all error-corrected chronograms. DaTeR uses a constrained optimization framework for the error-correction step, finding minimal deviations from previously assigned dates or branch lengths. We applied DaTeR to a biological dataset of 170 Cyanobacterial taxa and a reliable set of 24 transfer-based relative constraints, under six different molecular dating models. Our extensive analysis of this dataset demonstrates that DaTeR is both highly effective and scalable and that its application can significantly improve estimated chronograms.

   Freely available from https://compbio.engr.uconn.edu/software/dater/

   Supplementary data are available at Bioinformatics online.

    

   more » « less
5. 3D-equivariant graph neural networks for protein model quality assessment

   https://doi.org/10.1093/bioinformatics/btad030  

   Chen, Chen ; Chen, Xiao ; Morehead, Alex ; Wu, Tianqi ; Cheng, Jianlin ; Cowen, ed., Lenore ( January 2023 , Bioinformatics)

   Quality assessment (QA) of predicted protein tertiary structure models plays an important role in ranking and using them. With the recent development of deep learning end-to-end protein structure prediction techniques for generating highly confident tertiary structures for most proteins, it is important to explore corresponding QA strategies to evaluate and select the structural models predicted by them since these models have better quality and different properties than the models predicted by traditional tertiary structure prediction methods.

   We develop EnQA, a novel graph-based 3D-equivariant neural network method that is equivariant to rotation and translation of 3D objects to estimate the accuracy of protein structural models by leveraging the structural features acquired from the state-of-the-art tertiary structure prediction method—AlphaFold2. We train and test the method on both traditional model datasets (e.g. the datasets of the Critical Assessment of Techniques for Protein Structure Prediction) and a new dataset of high-quality structural models predicted only by AlphaFold2 for the proteins whose experimental structures were released recently. Our approach achieves state-of-the-art performance on protein structural models predicted by both traditional protein structure prediction methods and the latest end-to-end deep learning method—AlphaFold2. It performs even better than the model QA scores provided by AlphaFold2 itself. The results illustrate that the 3D-equivariant graph neural network is a promising approach to the evaluation of protein structural models. Integrating AlphaFold2 features with other complementary sequence and structural features is important for improving protein model QA.

   The source code is available at https://github.com/BioinfoMachineLearning/EnQA.

   Supplementary data are available at Bioinformatics online.

    

   more » « less