CpG Transformer for imputation of single-cell methylomes.
Ontology highlight
ABSTRACT: Motivation
The adoption of current single-cell DNA methylation sequencing protocols is hindered by incomplete coverage, outlining the need for effective imputation techniques. The task of imputing single-cell (methylation) data requires models to build an understanding of underlying biological processes.Results
We adapt the transformer neural network architecture to operate on methylation matrices through combining axial attention with sliding window self-attention. The obtained CpG Transformer displays state-of-the-art performances on a wide range of scBS-seq and scRRBS-seq datasets. Furthermore, we demonstrate the interpretability of CpG Transformer and illustrate its rapid transfer learning properties, allowing practitioners to train models on new datasets with a limited computational and time budget.Availability and implementation
CpG Transformer is freely available at https://github.com/gdewael/cpg-transformer.Supplementary information
Supplementary data are available at Bioinformatics online.
SUBMITTER: De Waele G
PROVIDER: S-EPMC8756163 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Measurements of single-cell methylation are revolutionizing our understanding of epigenetic control of gene expression, yet the intrinsic data sparsity limits the scope for quantitative analysis of such data. Here, we introduce Melissa (MEthyLation Inference for Single cell Analysis), a Bayesian hierarchical method to cluster cells based on local methylation patterns, discovering patterns of epigenetic variability between cells. The clustering also acts as an effective regularization for data imputation on unassayed CpG sites, enabling transfer of information between individual cells. We show both on simulated and real data sets that Melissa provides accurate and biologically meaningful clusterings and state-of-the-art imputation performance.
| S-EPMC6427844 | biostudies-literature
Project description:BACKGROUND:Whole-genome bisulfite sequencing (WGBS) has become the standard method for interrogating plant methylomes at base resolution. However, deep WGBS measurements remain cost prohibitive for large, complex genomes and for population-level studies. As a result, most published plant methylomes are sequenced far below saturation, with a large proportion of cytosines having either missing data or insufficient coverage. RESULTS:Here we present METHimpute, a Hidden Markov Model (HMM) based imputation algorithm for the analysis of WGBS data. Unlike existing methods, METHimpute enables the construction of complete methylomes by inferring the methylation status and level of all cytosines in the genome regardless of coverage. Application of METHimpute to maize, rice and Arabidopsis shows that the algorithm infers cytosine-resolution methylomes with high accuracy from data as low as 6X, compared to data with 60X, thus making it a cost-effective solution for large-scale studies. CONCLUSIONS:METHimpute provides methylation status calls and levels for all cytosines in the genome regardless of coverage, thus yielding complete methylomes even with low-coverage WGBS datasets. The method has been extensively tested in plants, but should also be applicable to other species. An implementation is available on Bioconductor.
| S-EPMC5992726 | biostudies-literature
Project description:Background: Single-cell RNA-seq is a powerful tool for measuring gene expression at the resolution of individual cells. A challenge in the analysis of this data is the large amount of zero values, representing either missing data or no expression. Several imputation approaches have been proposed to address this issue, but they generally rely on structure inherent to the dataset under consideration they may not provide any additional information, hence, are limited by the information contained therein and the validity of their assumptions. Methods: We evaluated the risk of generating false positive or irreproducible differential expression when imputing data with six different methods. We applied each method to a variety of simulated datasets as well as to permuted real single-cell RNA-seq datasets and consider the number of false positive gene-gene correlations and differentially expressed genes. Using matched 10X and Smart-seq2 data we examined whether cell-type specific markers were reproducible across datasets derived from the same tissue before and after imputation. Results: The extent of false-positives introduced by imputation varied considerably by method. Data smoothing based methods, MAGIC, knn-smooth and dca, generated many false-positives in both real and simulated data. Model-based imputation methods typically generated fewer false-positives but this varied greatly depending on the diversity of cell-types in the sample. All imputation methods decreased the reproducibility of cell-type specific markers, although this could be mitigated by selecting markers with large effect size and significance. Conclusions: Imputation of single-cell RNA-seq data introduces circularity that can generate false-positive results. Thus, statistical tests applied to imputed data should be treated with care. Additional filtering by effect size can reduce but not fully eliminate these effects. Of the methods we considered, SAVER was the least likely to generate false or irreproducible results, thus should be favoured over alternatives if imputation is necessary.
| S-EPMC6415334 | biostudies-literature
Project description:Single-cell DNA methylation has emerged as a powerful tool to study molecular heterogeneity within cellular populations. We report the development of a bisulfite based whole genome protocol suitable for use on single index sorted primary mammalian cells. Application of this protocol to human hematopoietic stem cells provided quantitative DNA methylation measurements of up to 5.7 million CpGs in a single cell. Analysis of cis CpG methylation states in single cells revealed a loss of concordance beyond 2kb, which supported the development of a new analytical framework (PDclust) for single-cell DNA methylation datasets that considers heterogeneity of methylation states at single CpG resolution. PDclust identified epigenetically distinct subpopulations within highly purified and functionally defined human stem cell populations. In silico merging of the methylation states from the single cells allowed us to ascribe function to these epigenetically distinct subpopulations on the basis of their epigenetic states.
| EGAS00001002789 | EGA
Project description:The mammalian brain contains diverse neuronal types, yet we lack single-cell epigenomic assays that are able to identify and characterize them. DNA methylation is a stable epigenetic mark that distinguishes cell types and marks regulatory elements. We generated >6000 methylomes from single neuronal nuclei and used them to identify 16 mouse and 21 human neuronal subpopulations in the frontal cortex. CG and non-CG methylation exhibited cell type-specific distributions, and we identified regulatory elements with differential methylation across neuron types. Methylation signatures identified a layer 6 excitatory neuron subtype and a unique human parvalbumin-expressing inhibitory neuron subtype. We observed stronger cross-species conservation of regulatory elements in inhibitory neurons than in excitatory neurons. Single-nucleus methylomes expand the atlas of brain cell types and identify regulatory elements that drive conserved brain cell diversity.
| S-EPMC5570439 | biostudies-literature
Project description:BACKGROUND:Single-cell RNA sequencing (scRNA-seq) technology provides an effective way to study cell heterogeneity. However, due to the low capture efficiency and stochastic gene expression, scRNA-seq data often contains a high percentage of missing values. It has been showed that the missing rate can reach approximately 30% even after noise reduction. To accurately recover missing values in scRNA-seq data, we need to know where the missing data is; how much data is missing; and what are the values of these data. METHODS:To solve these three problems, we propose a novel model with a hybrid machine learning method, namely, missing imputation for single-cell RNA-seq (MISC). To solve the first problem, we transformed it to a binary classification problem on the RNA-seq expression matrix. Then, for the second problem, we searched for the intersection of the classification results, zero-inflated model and false negative model results. Finally, we used the regression model to recover the data in the missing elements. RESULTS:We compared the raw data without imputation, the mean-smooth neighbor cell trajectory, MISC on chronic myeloid leukemia data (CML), the primary somatosensory cortex and the hippocampal CA1 region of mouse brain cells. On the CML data, MISC discovered a trajectory branch from the CP-CML to the BC-CML, which provides direct evidence of evolution from CP to BC stem cells. On the mouse brain data, MISC clearly divides the pyramidal CA1 into different branches, and it is direct evidence of pyramidal CA1 in the subpopulations. In the meantime, with MISC, the oligodendrocyte cells became an independent group with an apparent boundary. CONCLUSIONS:Our results showed that the MISC model improved the cell type classification and could be instrumental to study cellular heterogeneity. Overall, MISC is a robust missing data imputation model for single-cell RNA-seq data.
| S-EPMC6293493 | biostudies-literature
Project description:A key challenge in analyzing single cell RNA-sequencing data is the large number of false zeros, where genes actually expressed in a given cell are incorrectly measured as unexpressed. We present a method based on low-rank matrix approximation which imputes these values while preserving biologically non-expressed genes (true biological zeros) at zero expression levels. We provide theoretical justification for this denoising approach and demonstrate its advantages relative to other methods on simulated and biological datasets.
| S-EPMC8752663 | biostudies-literature
Project description:Cell type heterogeneity presents a challenge to the interpretation of epigenome data, compounded by the difficulty in generating reliable single-cell DNA methylomes for large numbers of cells and samples. We present EPISCORE, a computational algorithm that performs virtual microdissection of bulk tissue DNA methylation data at single cell-type resolution for any solid tissue. EPISCORE applies a probabilistic epigenetic model of gene regulation to a single-cell RNA-seq tissue atlas to generate a tissue-specific DNA methylation reference matrix, allowing quantification of cell-type proportions and cell-type-specific differential methylation signals in bulk tissue data. We validate EPISCORE in multiple epigenome studies and tissue types.
| S-EPMC7650528 | biostudies-literature
Project description:Studies of DNA methylomes in a wide range of eukaryotes have revealed both conserved and divergent characteristics of DNA methylation among phylogenetic groups. However, data on invertebrates particularly molluscs are limited, which hinders our understanding of the evolution of DNA methylation in metazoa. The sequencing of the Pacific oyster Crassostrea gigas genome provides an opportunity for genome-wide profiling of DNA methylation in this model mollusc.Homologous searches against the C. gigas genome identified functional orthologs for key genes involved in DNA methylation: DNMT1, DNMT2, DNMT3, MBD2/3 and UHRF1. Whole-genome bisulfite sequencing (BS-seq) of the oyster's mantle tissues revealed that more than 99% methylation modification was restricted to cytosines in CpG context and methylated CpGs accumulated in the bodies of genes that were moderately expressed. Young repeat elements were another major targets of CpG methylation in oysters. Comparison with other invertebrate methylomes suggested that the 5'-end bias of gene body methylation and the negative correlation between gene body methylation and gene length were the derived features probably limited to the insect lineage. Interestingly, phylostratigraphic analysis showed that CpG methylation preferentially targeted genes originating in the common ancestor of eukaryotes rather than the oldest genes originating in the common ancestor of cellular organisms.Comparative analysis of the oyster DNA methylomes and that of other animal species revealed that the characteristics of DNA methylation were generally conserved during invertebrate evolution, while some unique features were derived in the insect lineage. The preference of methylation modification on genes originating in the eukaryotic ancestor rather than the oldest genes is unexpected, probably implying that the emergence of methylation regulation in these 'relatively young' genes was critical for the origin and radiation of eukaryotes.
| S-EPMC4378273 | biostudies-literature
Project description:The emergence of single-cell RNA sequencing (scRNA-seq) technologies has enabled us to measure the expression levels of thousands of genes at single-cell resolution. However, insufficient quantities of starting RNA in the individual cells cause significant dropout events, introducing a large number of zero counts in the expression matrix. To circumvent this, we developed an autoencoder-based sparse gene expression matrix imputation method. AutoImpute, which learns the inherent distribution of the input scRNA-seq data and imputes the missing values accordingly with minimal modification to the biologically silent genes. When tested on real scRNA-seq datasets, AutoImpute performed competitively wrt., the existing single-cell imputation methods, on the grounds of expression recovery from subsampled data, cell-clustering accuracy, variance stabilization and cell-type separability.
| S-EPMC6218547 | biostudies-literature