ncibtep@nih.gov

Bioinformatics Training and Education Program

TOTM – November 2017: Single Cell RNA-Seq on your mind? Read this before starting your adventure!

[et_pb_section bb_built=”1″ admin_label=”section” _builder_version=”3.0.86″ background_blend=”exclusion” background_size=”contain” background_repeat=”repeat-y” use_background_color_gradient=”on” background_color_gradient_start=”rgba(252,49,42,0.81)” background_color_gradient_end=”rgba(255,255,255,0.73)” background_color_gradient_type=”radial”][et_pb_row admin_label=”row” background_size=”contain” _builder_version=”3.0.86″ background_blend=”exclusion” background_repeat=”repeat-x”][et_pb_column type=”4_4″][et_pb_text background_position=”top_left” background_repeat=”repeat” background_size=”initial” _builder_version=”3.0.85″ background_layout=”light”]

If you are considering single cell RNA-Seq (scRNAseq) as a method to generate meaningful information for your research project, here are 7 questions to ask yourself right now, PRIOR to doing any ‘quick-and-dirty’ pilot study.

  1. Is single cell RNA-Seq the answer? Single cell RNA-Seq tends to be more expensive, prone to more technical noise, less sensitive for detection of expression, and generally does not preserve spatial information of the cells assayed. While it is a powerful technique that produces a flexible data set, a good starting point when considering a project is to have a clear reason why you might want or need single cell RNA-Seq. This usually includes a desire to assay a heterogeneous population, a dynamic process, or taking a survey of various cell types within a larger tissue or system.
  2. What will you ask from your data? Once you decide single cell RNA-Seq is likely the approach you want to use for your goals, it is important to make sure you have an appropriate experimental design. As in bulk RNA-Seq, identification of genes that are enriched in certain cells or under certain conditions require something to which you will compare expression and proper controls. These design considerations can sometimes be internal to your experiment, but even a basic survey of cells and the genes that they express is aided by thoughtful design and good controls. What is the effect of animal to animal difference, effect of gender, etc? Variation can come from many sources – increase your chances of detecting what you set out to measure by being able to account for other sources of variation in your experiment.
  3. Are you able to get single cells? Most single cell RNA-Seq protocols require dissociated, viable cells to be used as input for downstream steps such as lysis, RNA reverse transcription and sample barcoding. Some tissues are difficult to isolate cells from, some cells are very fragile, and evidence suggests that the process of preparing cells can result in a change to the transcriptome of the cells. There are some promising approaches to work around some of these issues, but it is good idea to consider how you are going to isolate your cells and have a sense of what the advantage and disadvantage are of some of the options.
  4. Which method should you use? Numerous methods exist for assaying expression in single cells. They differ in their strengths and limitations in terms of cost per single cell sample, sensitivity of detection, throughput, requirement of specialized equipment and whether full-length transcript information is generated. The choice of which method to use will often depend on the specific questions asked in the research project. It may help to talk with someone who has knowledge of the various options, as well as those who will be involved in the bioinformatic analysis (since the type of data generated will influence the analysis involved).
  5. How many cells do you need? This may be one of the more difficult questions to answer until you generate some initial data. It depends on many aspects, including your experiment design, the single cell RNA-Seq method you selected, the complexity of your sample, and the expression level of genes within your sample. There are some basic methods and tools available for doing a power analysis to give you a range, but most require a sense of the basic features of your sample such as percentage of target cells within the total population and relative expression level of key genes. Existing histology or flow cytometry data may help with this.
  6. Who will analyze the data? Single cell RNA-Seq generates a highly-multidimensional dataset with the expression of many genes assayed across many cells. Not surprisingly, this creates some exciting possibilities in terms of the types of analysis that can be done, but also adds some extra computational and statistical complexity. Single cell RNA-Seq analysis also differs somewhat from bulk RNA-Seq in that some of the underlying statistical assumptions are different, and some types of analysis leverage biological subject matter expertise. It is a good idea to identify who will be involved in analyzing the data early on and include them at the experimental design stage and initial selection and evaluation of methods.
  7. Data analyzed; what’s next? Whether part of a larger study, meant as a survey to find candidates, or as the generation of a gene expression resource, the confidence in and utility of the data can be supported by validation. For well-characterized cells, existing literature and data may work, but one should consider confirming novel claims. Confirmation of cell localization, co-expression of identified genes, and/or showing associated protein expression are some of the more common forms of validation of single cell RNA-Seq datasets. For projects where single cell RNA-Seq is used to identify more functional or dynamic biological processes, a more mechanistic form of validation may help support the analysis. A rigorously collected and validated single cell RNA-Seq dataset may be useful far beyond answering the original research questions.

REFERENCES

  1. Haque, A., Engel, J., Teichmann, S. A. & Lönnberg, T. A practical guide to single-cell RNA-sequencing for biomedical research and clinical applications. Genome Med. 9, 75 (2017). [Note from the author of the blog: this article is a “must-read” practical guide to single-cell RNA-Seq. Link to Article
  2. Stegle, O., Teichmann, S. A. & Marioni, J. C. Computational and analytical challenges in single-cell transcriptomics. Nat Rev Genet 16, 133–145 (2015).
  3. Poulin, J.-F., Tasic, B., Hjerling-Leffler, J., Trimarchi, J. M. & Awatramani, R. Disentangling neural cell diversity using single-cell transcriptomics. Neurosci. 19, 1131–1141 (2016).
  4. Liu, Serena, and Cole Trapnell. “Single-Cell Transcriptome Sequencing: Recent Advances and Remaining Challenges.” F1000Research 5 (2016): F1000 Faculty Rev–182. PMC. 6 Nov. 2017.
  5. Zhu, X., Wolfgruber, T., Tasato, A., Garmire, D. & Garmire, L. X. Granatum: a graphical single-cell RNA-Seq analysis pipeline for genomics scientists. bioRxiv (2017). Preprint link: https://www.biorxiv.org/content/early/2017/08/08/110759; accepted in Genome Med. November 2017.

 

Michael Kelly, Ph.D. (Laboratory of Cochlear Development (NIDCD)), is a single cell genomics expert with a broad knowledge of the field, and with a track record for successfully implementing best practices for single cell gene expression analysis. He has consulted and collaborated with many intramural and extramural laboratories on experimental design, method selection and use, and analysis of data for single cell genomics studies. Mike enthusiastically promotes an open and collaborative community of researchers to advance single cell biology at NIH, through his work on the Single Cell Genomics SIG steering committee and Single Cell Users Group activities.

[/et_pb_text][/et_pb_column][/et_pb_row][/et_pb_section]