How many reads do I need for RNA-seq? How do I choose the sequencing depth?

For standard differential-expression analysis of genes using mRNA-seq, it is usually sufficient to generate approximately 20-30 million read pairs per sample. Greater depth may be needed when analysing lowly expressed genes, novel transcripts, alternative splicing, mixed material or Total RNA-seq libraries.

However, there is no single correct value for every project. You should choose the sequencing depth according to the analysis objective, library type, transcriptome complexity, material quality and number of biological replicates.

Comparison of RNA-seq sequencing depth options

The values below apply to paired-end 2 × 150 bp libraries.

These values are indicative. The actual number of usable reads depends on sequencing quality, rRNA removal efficiency, the duplication rate and the number of reads that map uniquely to the genome or transcriptome.

Reads or read pairs?

In paired-end projects, each library fragment is read from both ends. Sequencing 20 million fragments in 2 × 150 bp mode therefore generates:

• 20 million read pairs,
• 40 million individual reads,
• approximately 6 Gb of raw data.

The term '40 million reads' can therefore mean either 40 million individual reads or 40 million read pairs. When describing a project for a quotation, always specify whether the value refers to individual reads, read pairs or the total number of bases.

When should you choose 6 Gb - approximately 20 million read pairs?

The 6 Gb option is suitable if:

• you are performing standard mRNA-seq with Poly(A) selection,
• the main aim is differential-expression analysis of genes,
• the organism under study has a high-quality genome and annotation,
• the samples come from a homogeneous tissue or cell population,
• RNA is of good quality,
• you are not planning an extensive analysis of novel isoforms,
• the genes being analysed are not expressed at extremely low levels.

This option is often sufficient for standard experiments comparing gene expression between conditions, tissues, genotypes or time points.

When should you choose 10 Gb - approximately 33 million read pairs?

The 10 Gb option is worth considering if:

• you want greater detection sensitivity for lowly expressed genes,
• you are analysing a more complex transcriptome,
• the material includes several cell types,
• alternative splicing analysis is planned,
• you expect moderately large differences in gene expression,
• the genome or its annotation is incomplete,
• you want to allow a greater margin for discarding reads during quality control.

This option offers a reasonable compromise between cost and the breadth of information obtained in more demanding mRNA-seq projects.

When should you choose 15 Gb - approximately 50 million read pairs?

The 15 Gb option is justified if:

• you plan to analyse novel transcripts and isoforms,
• you are interested in differential exon usage,
• you are analysing complex alternative splicing,
• you are looking for lowly expressed transcripts,
• the samples are heterogeneous,
• the transcriptome of the organism under study is large or poorly characterised,
• it is important to analyse variants using data from RNA-seq,
• you expect a lower proportion of usable reads.

Larger amount of data does not, however, guarantee the detection of all transcripts. Library complexity, RNA quality or very low expression levels may be limiting factors.

When should you choose 12-16 Gb for Total RNA-seq?

Total RNA-seq with rRNA depletion usually requires a greater amount of data than mRNA-seq with Poly(A) selection. After rRNA removal, the library still contains various types of RNA, including mRNA, lncRNA, antisense transcripts, pre-mRNA and other non-coding RNAs.

Choose approximately 12 Gb per sample if:

• you are performing standard Total RNA-seq,
• you are analysing mRNA and lncRNA,
• you are interested in transcripts without a poly(A) tail,
• rRNA depletion efficiency is expected to be high.

Choose approximately 16 Gb or more if:

• the project includes circRNA analysis,
• you are looking for rare non-coding transcripts,
• the material is highly complex,
• you expect a high proportion of residual rRNA,
• the sample contains RNA from several organisms.

To analyse mature miRNAs and other very short RNAs, use a separate protocol for small RNA-seq.

When will greater depth not solve the problem?

Additional reads help only if the library contains enough distinct molecules. If library complexity is low, further sequencing will mainly generate duplicate reads.

This applies particularly to:

• low-input samples,
• RNA that is highly degraded,
• libraries subjected to extensive PCR amplification,
• material with a very low cell count,
• samples with ineffective rRNA depletion,
• libraries with an abnormal fragment size distribution.

In such cases, material quality and the library preparation method are more important than simply increasing the number of reads.

Replicates or more reads?

In differential-expression analysis, the number of independent biological replicates is often more important than sequencing a small number of samples at very high depth.

If the budget is limited, it is usually better to:

• include at least three independent biological replicates per group,
• use a moderate but sufficient depth,
• maintain a balanced experimental design,
• avoid confounding a biological group with a specific library preparation or sequencing batch.

Increasing sequencing depth for the same sample does not replace biological replication and does not allow you to estimate inter-individual variability.

Sample Requirements

mRNA-seq with Poly(A) selection

Recommended parameters:

• material: total RNA,
• amount: above 0.5 µg,
• concentration: above 50 ng/µl,
• volume: above 20 µl,
• OD260/280: 1.8-2.2,
• RIN: above 7, or above 6 for plants.

Minimum parameters:

• amount: above 0.2 µg,
• concentration measured using Qubit: above 20 ng/µl,
• volume: above 20 µl,
• OD260/280: 1.7-2.5,
• RIN: above 4.5, or above 4 for plants.

Total RNA-seq with rRNA depletion

Recommended parameters:

• material: total RNA,
• amount: above 2 µg,
• concentration: above 100 ng/µl,
• volume: above 10 µl,
• OD260/280: 1.8-2.2,
• RIN: above 7, or above 6 for plants.

For low-input projects, FACS-sorted samples, FFPE material or RNA that does not meet standard requirements, the parameters should be assessed on a case-by-case basis. A smaller amount of RNA may require a change in protocol, but does not always mean that sequencing depth should be increased.

What affects the number of reads needed?

Common mistakes when planning sequencing depth

Project planning questions

Before choosing the amount of data, consider these questions:

Plan your project

Describe your project - we will recommend a method and amount of data.

Based on the library type, the organism under study, RNA quality and the planned scope of analysis, we will help you determine the number of reads needed for each sample. We will also select the library preparation method, sequencing mode and scope of bioinformatics analysis.