Single Cell Journal Club blog
The Single Cell Multimodal Journal Club parses the emerging literature that ranges from reviews, experimental innovations and analysis pipeline papers. The group meets weekly on Zoom and consists of both bioinformaticians and biologists. This blog is a casual reflection of the papers and our discussion.
“The secret life of cells” by Michael Eisenstein was our first paper, an editorial piece with a coy and curious title. As if cells have an agenda to hide their molecular business. What kind of business does a cell have anyways? Cellular business generally entails receiving, processing and responding to stimuli. Researchers have been trying to decode the processing language and its mechanisms since the discovery of cells in 1665 by Robert Hooke. With his candle lit microscope, Hooke described individual ‘miniaturized rooms’, which we know today as single cells. Single cell technology has evolved far from Hooke’s makeshift microscope, yet the desire to unveil the secrets of cells remains.
Eisenstein’s paper highlights emerging single cell multimodal methods and the people behind them. Instead of re-writing the paper, I will provide a table summary of the sequencing method, the scientist (I am crediting the people mentioned in the paper only. Please forgive me for missing the teams involved), the type of molecules measured and the type of experiments:
| Method | Person | Molecules measured | Case studies |
|---|---|---|---|
| scG&T | Theirry Voet | DNA and mRNA | Measure genetic variation and transcription, DNA-based cell-lineage trees |
| scATAC-seq | William Greenleaf | Chromatin accessibility and mRNA | Measure gene regulation, Differentiate cell states |
| scM&T | He, Steagle, Reik, and Kelsey | Methylation and mRNA | Track epigenetic marks, Cell lineage |
| SNARE-seq | Kun Zhang | scRNA-seq of nuclei | scRNA-seq of nuclei from bulk tissue |
| sci-CAR | Cole Trapnell, Jay Shendure | scRNA-seq of nuclei using combinatorial indexing | scRNA-seq of nuclei from bulk tissue |
| CITE-seq | Rahul Satija | Cell surface proteins and mRNA | Detect subclonal population (ex: t-cell heterogeneity) |
| ECCITE-seq | Rahul Satija | Cell surface proteins and mRNA | sgRNA (single-guided RNA) allows for tracking CAS9 mutations |
| epiScanpy | Maria Colomé-Tatche | Integrate DNA methylation, Chromatin accessibility and mRNA | Statistical single cell multi-omics integration, Epigenetic marks analysis platform |
Science, Cao, J., Cusanovich, D. A., Ramani, V., Aghamirzaie, D., Pliner, H. A., Hill, A. J., … Shendure, J. (2018) doi:10.1126/science.aau0730
A 2018 proof of concept paper on a new single cell co-assay experiment method: sci-CAR (single cell chromatin accessibility and RNA-seq). sci-CAR allows for simultaneous measurements of mRNA and chromatin accessibility from ~16,000 nuclei. How can sci-CAR measure nearly 1000X more cells than other single cell multimodal methods? Combinatorial indexing.
What makes sci-CAR so powerful in comparison to other single cell platforms is the combinatorial indexing method. Combinatorial indexing enables tracking of individual sequence molecules (in this case: mRNA and Tn5 DNA tags) to the same cell because of unique molecular identifier (UMI) adapters. UMIs are nucleotide sequences that act as barcodes. UMIs in are introduced in both the 3’ poly-A tail of mRNAs amplification and during tagmentation (tagmentation is when a transposon cleaves and adds a nucleotide sequence tag on double-stranded DNA) process of Tn5.
In short, combinatorial indexing molecular material allows for integrated analysis across modalities. sci-CAR is able to accurately correlate RNA-seq data with ATAC-seq to the correct organism in a barnyard quality control experiment (mixing human and mouse cells together). Initial efforts to cluster sci-ATAC-seq profiles failed to discover the expected diversity of cell types, likely due to the sparsity of the data. But, by clustering cells based on RNA expression profiles and aggregated chromatin accessibility profiles uncovered 22,000 regulatory elements. And, by using covariance of differential RNA expression and differential chromatin accessibility, Cao et al. linked new distal elements to gene expression. Sometimes, chromatin accessibility correlated better with a distal element than its own promoter. Links between distal cis-regulatory elements and their target genes can be useful for explaining differential expression across cell types
Our group had comments and questions on the 3 main types of points, either experimental, analysis, and biological.
Experimental
-
Only a subset of the nuclei have both modalities. Maybe linked to the extreme sparsity? Both modalities have similar missingness. Does splitting the lysate contribute to sparsity?
-
The architecture of the genome is important to consider when thinking about technical biases, like a short gene might be entirely exposed, but a long gene is exposed in components.
-
Sequencing depth vs number of cells: design question. With ATAC, you can only get two reads per loci per cell (2 chromosomes), so you need a lot of cells for replication and peak calling (is a peak where you see two overlapping reads?). May be partially responsible for low correlation between ATAC and RNA-seq?
Analysis
-
Complexity of ATAC-seq library compared to single-cell unimodal and bulk ATAC-seq: shallow sequencing is problematic, what is a peak if you only have 7000 reads in the whole genome?
-
Is there any kind of “integrated” toy data set for this method, i.e. a common ground for method development.
-
Is there standardized software for downstream analysis? But most of the work is done before the experiment, to prep the indexing methods.
Biological
-
Gene regulation network including the epigenome information. Would this data be a good candidate, maybe not because of the lack of overlap between modalities.
-
Does the accessibility and expression always need to be differential?
Although few sci-CAR datasets are currently available, there is consensus that combinatorial indexing will continue to be at the frontiers of single cell multimodal methodologies. The single cell sequencing field is moving quickly, so stay tuned for more SCMM journal club summaries!