SCSIM: Jointly simulating correlated single-cell and bulk next-generation DNA sequencing data

Giguere, Collin; Dubey, Harsh Vardhan; Sarsani, Vishal Kumar; Saddiki, Hachem; He, Shai; Flaherty, Patrick

doi:10.1101/2020.02.03.930354

Cited by 3 publications

(4 citation statements)

References 12 publications

(14 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For scalability analysis, simulated datasets are generated using the scsim Python package 56 , which is based on the Splatter statistical framework while it performs better efficiency to generate large scale simulated data. Following parameters are used in scsim() Python function: ngenes=25000, ncells=50000, 100000, 50000, 1000000, 2000000, 10000000, ngroups=5, diffexpprob= 0.025.…”

Section: Scalability Analysismentioning

confidence: 99%

Self-distillation contrastive learning enables clustering-free signature extraction and mapping to multimodal single-cell atlas of multimillion scale

Yang

Huang

et al. 2021

Preprint

View full text Add to dashboard Cite

Massively generated single-cell multi-omics datasets are revolutionizing biological studies of heterogenous tissues and organisms, which necessitate powerful computational methods to unleash the full potential of these tremendous data. Here, we present Concerto, stands for self-distillation contrastive learning of cell representations, a self-supervised representation learning framework optimized with asymmetric teacher-student configuration to analyze single-cell multi-omics datasets with scalability up to building 10 million-cell reference within 1.5 hour and querying 10k cells within 8 seconds. Concerto leverages dropout layer as minimal data augmentation to learn meaningful cell representations in a contrastive manner. The teacher module uses attention mechanism to aggregate contextualized gene embeddings within cellular context, while the student module uses simpler dense structure with discreate input. The learned task-agnostic representations can be adapted to a broad range of single-cell computation tasks. 1) Via supervised fine-tuning, Concerto enables automatic cell classification as well as novel cell-type discovery; 2) Attention weights provide model interpretability via automatically extracting specific molecular signatures at single-cell resolution without the needs of clustering; 3) Via source-aware training, Concerto supports efficient data integration by projecting all cells across multiple batches into a joint embedding space. 4) Via batch-aware inference or unsupervised fine-tuning, Concerto enables mapping query cells onto reference and accurately transferring annotations. Concerto can flexibly extend to multi-omics datasets simply through cross-modality summation operation to obtain unified cell embeddings. Using examples from human peripheral blood, human thymus, human pancreas, and mouse tissue atlas, Concerto shows superior performance benchmarking against other top-performing methods. We also demonstrate Concerto recapitulates detailed COVID-19 disease variation through query-to-reference mapping. Concerto can operate on all genes and represents a fully data-driven approach with minimum prior distribution assumptions, eliminating the needs of PCA-like or autoencoder-like dimensionality reduction, which significantly reforms the current best practice. Concerto is a simple, straightforward, robust, and scalable framework, offering a brand new perspective to derive cell representations and can effectively satisfy the emerging paradigm of query-to-reference mapping in the era of atlas-level single-cell multimodal analysis.

show abstract

Section: Scalability Analysismentioning

confidence: 99%

Self-distillation contrastive learning enables clustering-free signature extraction and mapping to multimodal single-cell atlas of multimillion scale

Yang

Huang

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…There have been various read simulators built for single-cell genomic sequencing data. SimSCScTree[16], SCSIM[17], SCSsim[18] and CellCoal[19] are able to generate sequencing reads under single-cell resolution to benchmark analysis tools for single nucleotide variants(SNVs)[20] or copy number aberrations (CNAs)[21]. In contrast, few methods can simulate sequencing reads for single-cell epigenomic and transcriptomic data.…”

Section: Introductionmentioning

confidence: 99%

scReadSim: a single-cell RNA-seq and ATAC-seq read simulator

Yan

2022

Preprint

View full text Add to dashboard Cite

Single-cell sequencing technologies emerged and diversified rapidly in the past few years, along with the successful development of many computational tools. Realistic simulators can help researchers benchmark computational tools. However, few simulators can generate single-cell multi-omics data, and none can generate reads directly. To fill in this gap, we propose scReadSim, a simulator for single-cell multi-omics reads. Trained on real data, scReadSim generates synthetic sequencing reads in BAM or FASTQ formats. We deployed scReadSim on a sci-ATAC-seq dataset and a single-cell multimodal dataset to show the resemblance between synthetic and real data at the read and count levels. Moreover, we show that scReadSim allows user-specified ground truths of accessible chromatin regions for single-cell chromatin accessibility data generation. In addition, scReadSim is flexible for allowing varying throughputs and library sizes as input parameters to guide experimental design.

show abstract

“…Yu et al developed SCSsim to produce SNVs, Indels, and CNVs, especially tackling the issue of allele dropout (ADO) and alleles unbalanced amplification frequently occurs in scDNA-seq [ 20 ]. SCSIM jointly mimics correlated single-cell and bulk DNA reads with SNVs [ 21 ]. Mallory et al developed SingleCellCNABenchmark which generates in silico single-cell reads with CNVs [ 22 ].…”

Section: Introductionmentioning

confidence: 99%

“…Evol. - - - - SCSsim [ 20 ] Bioinformatics - - SCSIM [ 21 ] BMC Bioinformatics - - - - SingleCellCNABenchmark [ 22 ] PLoS Comput. Biol - - - - SCSilicon - …”

Section: Introductionmentioning

confidence: 99%

SCSilicon: a tool for synthetic single-cell DNA sequencing data generation

Feng

Chen

2022

BMC Genomics

View full text Add to dashboard Cite

Background Single-cell DNA sequencing is getting indispensable in the study of cell-specific cancer genomics. The performance of computational tools that tackle single-cell genome aberrations may be nevertheless undervalued or overvalued, owing to the insufficient size of benchmarking data. In silicon simulation is a cost-effective approach to generate as many single-cell genomes as possible in a controlled manner to make reliable and valid benchmarking. Results This study proposes a new tool, SCSilicon, which efficiently generates single-cell in silicon DNA reads with minimum manual intervention. SCSilicon automatically creates a set of genomic aberrations, including SNP, SNV, Indel, and CNV. Besides, SCSilicon yields the ground truth of CNV segmentation breakpoints and subclone cell labels. We have manually inspected a series of synthetic variations. We conducted a sanity check of the start-of-the-art single-cell CNV callers and found SCYN was the most robust one. Conclusions SCSilicon is a user-friendly software package for users to develop and benchmark single-cell CNV callers. Source code of SCSilicon is available at https://github.com/xikanfeng2/SCSilicon.

show abstract

SCSIM: Jointly simulating correlated single-cell and bulk next-generation DNA sequencing data

Cited by 3 publications

References 12 publications

Self-distillation contrastive learning enables clustering-free signature extraction and mapping to multimodal single-cell atlas of multimillion scale

Self-distillation contrastive learning enables clustering-free signature extraction and mapping to multimodal single-cell atlas of multimillion scale

scReadSim: a single-cell RNA-seq and ATAC-seq read simulator

SCSilicon: a tool for synthetic single-cell DNA sequencing data generation

Contact Info

Product

Resources

About