Approximate time: 50 minutes

Learning Objectives:

Experimental planning considerations

Understanding the steps in the experimental process of RNA extraction and preparation of RNA-Seq libraries is helpful for designing an RNA-Seq experiment, but there are special considerations that should be highlighted that can greatly affect the quality of a differential expression analysis.

These important considerations include:

  1. Number and type of replicates
  2. Avoiding confounding
  3. Addressing batch effects

We will go over each of these considerations in detail, discussing best practice and optimal design.


Experimental replicates can be performed as technical replicates or biological replicates.

Image credit: Klaus B., EMBO J (2015) 34: 2727-2730

In the days of microarrays, technical replicates were considered a necessity; however, with the current RNA-Seq technologies, technical variation is much lower than biological variation and technical replicates are unneccessary.

In contrast, biological replicates are absolutely essential for differential expression analysis. For mice or rats, this might be easy to determine what constitutes a different biological sample, but it’s a bit more difficult to determine for cell lines. This article gives some great recommendations for cell line replicates.

For differential expression analysis, the more biological replicates, the better the estimates of biological variation and the more precise our estimates of the mean expression levels. This leads to more accurate modeling of our data and identification of more differentially expressed genes.

Image credit: Liu, Y., et al., Bioinformatics (2014) 30(3): 301–304

As the figure above illustrates, biological replicates are of greater importance than sequencing depth, which is the total number of reads sequenced per sample. The figure shows the relationship between sequencing depth and number of replicates on the number of differentially expressed genes identified [1]. Note that an increase in the number of replicates tends to return more DE genes than increasing the sequencing depth. Therefore, generally more replicates are better than higher sequencing depth, with the caveat that higher depth is required for detection of lowly expressed DE genes and for performing isoform-level differential expression.

Sample pooling: Try to avoid pooling of individuals/experiments, if possible; however, if absolutely necessary, then each pooled set of samples would count as a single replicate. To ensure similar amounts of variation between replicates, you would want to pool the same number of individuals for each pooled set of samples.

For example, if you need at least 3 individuals to get enough material for your control replicate and at least 5 individuals to get enough material for your treatment replicate, you would pool 5 individuals for the control and 5 individuals for the treatment conditions. You would also make sure that the individuals that are pooled in both conditions are similar in sex, age, etc.

Replicates are almost always preferred to greater sequencing depth for bulk RNA-Seq. However, guidelines depend on the experiment performed and the desired analysis. Below we list some general guidelines for replicates and sequencing depth to help with experimental planning:

NOTE: The factor used to estimate the depth of sequencing for genomes is “coverage” - how many times do the number of nucleotides sequenced “cover” the genome. This metric is not exact for genomes (whole genome sequencing), but it is good enough and is used extensively. However, the metric does not work for transcriptomes because even though you may know what % of the genome has trancriptional activity, the expression of the genes is highly variable.


A confounded RNA-Seq experiment is one where you cannot distinguish the separate effects of two different sources of variation in the data.

For example, we know that sex has large effects on gene expression, and if all of our control mice were female and all of the treatment mice were male, then our treatment effect would be confounded by sex. We could not differentiate the effect of treatment from the effect of sex.

To AVOID confounding:

Batch effects

Batch effects are a significant issue for RNA-Seq analyses, since you can see significant differences in expression due solely to the batch effect.

Image credit: Hicks SC, et al., bioRxiv (2015)

To explore the issues generated by poor batch study design, they are highlighted nicely in this paper.

How to know whether you have batches?

If any of the answers is ‘No’, then you have batches.

Best practices regarding batches:

NOTE: The sample preparation of cell line “biological” replicates “should be performed as independently as possible” (as batches), “meaning that cell culture media should be prepared freshly for each experiment, different frozen cell stocks and growth factor batches, etc. should be used [2].” However, preparation across all conditions should be performed at the same time.


Your experiment has three different treatment groups, A, B, and C. Due to the lengthy process of tissue extraction, you can only isolate the RNA from two samples at the same time. You plan to have 4 replicates per group.

  1. Fill in the RNA isolation column of the metadata table. Since we can only prepare 2 samples at a time and we have 12 samples total, you will need to isolate RNA in 6 batches. In the RNA isolation column, enter one of the following values for each sample: group1, group2, group3, group4, group5, group6. Make sure to fill in the table so as to avoid confounding by batch of RNA isolation.

    Click here to download the below table as an Excel file.

  2. BONUS: To perform the RNA isolations more quickly, you devote two researchers to perform the RNA isolations. Create a researcher column and fill in the researchers’ initials for the samples they will prepare: use initials AB or CD.

sample treatment sex replicate RNA isolation
sample1 A F 1  
sample2 A F 2  
sample3 A M 3  
sample4 A M 4  
sample5 B F 1  
sample6 B F 2  
sample7 B M 3  
sample8 B M 4  
sample9 C F 1  
sample10 C F 2  
sample11 C M 3  
sample12 C M 4  

This lesson has been developed by members of the teaching team at the Harvard Chan Bioinformatics Core (HBC). These are open access materials distributed under the terms of the Creative Commons Attribution license (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.