A Beginner's Guide to RNA-Seq Analysis

This post serves as a comprehensive guide to RNA-Seq analysis, highlighting the tools and approaches.

Understanding RNA-Seq's Place in the Research Landscape

RNA-Seq is a powerful tool for studying the link between DNA sequence and phenotype. It provides quantitative data about gene expression, allowing researchers to trace the path from genetic variants identified in QTL analyses to their impact on behavioral or physiological traits.

This post focuses on this crucial area, specifically on how to move from a genomic region of interest to identifying genes involved in the observed phenotype. RNA expression serves as a valuable bridge in this process.

Outline:

• Why RNA-Seq? – Understanding the significance of RNA-Seq as a quantitative tool for gene expression analysis.
• Technical Overview of Illumina Sequencing – A detailed overview of the Illumina method for sequencing, one of the most widely used technologies in the field.
• Considerations for Experimental Design – Key factors to consider when designing an RNA-Seq experiment, including RNA fraction, library type, sequencing length, depth, and number of replicates.
• From Reads to Quantitative Expression Measurements – The steps involved in converting RNA-Seq reads into quantifiable expression measurements and the use of common differential expression analysis tools.

Why RNA-Seq?

RNA-Seq has revolutionized gene expression analysis, offering several advantages over traditional methods:

• Quantitative link between DNA and phenotype: RNA-Seq provides the first quantitative connection between DNA sequence and observable traits, making it crucial for understanding the functional implications of genetic variations.
• Understanding the full transcriptome: RNA-Seq goes beyond protein-coding genes to provide data on a variety of non-coding RNAs, including microRNAs, long non-coding RNAs, snoRNAs, and circular RNAs, offering a more complete picture of gene expression.
• Discovery of new transcripts: RNA-Seq enables the identification of novel protein-coding genes, splice variants, and non-coding transcripts, expanding our knowledge of the transcriptome.

Technical Overview of Illumina Sequencing

Illumina sequencing is a widely adopted technology for RNA-Seq. Here's a simplified overview of the process:

1. RNA extraction: Start with a sample of interest and isolate the RNA.
2. RNA fraction selection: Choose the RNA fraction you are most interested in, whether it's mRNA, total RNA, or small RNAs.
3. Library preparation: The RNA is fragmented, adapters are added, and the RNA is converted to cDNA for sequencing.
4. Sequencing: The library is loaded onto a flow cell and sequenced using a high-throughput sequencing system like HiSeq, NovaSeq, or MiSeq.
5. Read analysis: The sequenced reads are analyzed to obtain information about gene expression.

A Visual Guide: For a visual explanation of the Illumina sequencing process, check out this informative YouTube video from Illumina.

Considerations for Experimental Design

Here are key aspects to consider when planning your RNA-Seq experiment:

RNA Fraction:

• PolyA selection: This is the traditional approach, targeting protein-coding genes by capturing transcripts with polyA tails.
• Ribosomal RNA depletion: This method removes ribosomal RNA, allowing for a broader measurement of the transcriptome, including long non-coding RNAs.
• Size selection: This method isolates specific size ranges of RNA, often used to enrich for small RNAs like microRNAs.

Library Type:

• Single-end reads: A single read is obtained from each RNA fragment.
• Paired-end reads: Reads are obtained from both ends of each RNA fragment, providing more information on transcript structure and alignment.
• Stranded versus unstranded libraries: Stranded libraries provide information on the strand of origin for each transcript, allowing for more accurate mapping of overlapping transcripts. Unstranded libraries do not distinguish between strands.

Sequencing Length:

• Longer reads (e.g., 150 base pairs) improve mapping accuracy and transcript identification but are more expensive.
• Shorter reads (e.g., 50 base pairs) are sufficient for gene-level analysis and are cost-effective.

Sequencing Depth:

• Sequencing depth refers to the number of reads per sample.
• Deeper sequencing provides more precise estimates for a greater number of genes.
• Recommended depths vary depending on the specific research question, but a general rule of thumb for gene-level analysis is 20-30 million reads per sample, with higher depths needed for isoform-level analysis or poorly annotated transcriptomes.

Number of Replicates:

• The optimal number of replicates depends on factors like group homogeneity, desired power, and the variance within the groups.
• A general guideline for inbred strains for gene-level analysis is four replicates per group.
• More heterogeneous groups or treatment groups require more replicates.

Other considerations:

• Synthetic spike-ins: These are external transcripts added to the sample to assess technical variability and enable absolute quantification.
• Library preparation randomization: Randomize samples during library preparation to minimize batch effects.
• Multiplexing: Pool samples with unique tags to allow for sequencing on multiple lanes, reducing batch effects and increasing efficiency.

From Reads to Quantitative Expression Measurements

This section focuses on the steps involved in analyzing RNA-Seq data to generate quantifiable gene expression measurements:

1. Quality Control of Raw Reads:

• Ensure the correct number of reads and read length for each sample.
• Use tools like FastQC to assess read quality and identify potential issues like poor quality base calls or adapter contamination.
• Trimming can be used to remove adapter sequences and low-quality base calls, providing valuable insights into RNA degradation or sample integrity.

2. Alignment to the Genome (Optional):

• Aligning reads to the genome is crucial for de novo transcriptome assembly but can be bypassed if a well-annotated transcriptome is available.
• Tools like HISAT2 are used for aligning reads, taking into account splice junctions.
• Alignment metrics like the number of concordantly aligned reads provide insights into alignment quality.

3. Transcript Discovery (Optional):

• Transcript discovery is valuable for studying splicing and non-coding RNAs, particularly when the transcriptome is poorly annotated.
• Tools like StringTie are used to reconstruct transcripts based on read alignments and coverage.
• The resulting GTF file can be visualized in genome browsers like UCSC Genome Browser to identify novel transcripts and splice variants.

4. Alignment to the Transcriptome and Quantification:

• Aligning reads to a known transcriptome (either a reference transcriptome or a de novo assembled one) allows for quantification of gene expression.
• Tools like RSEM are used for quantifying gene and isoform expression based on read alignments.
• RSEM uses an expectation-maximization algorithm to estimate read counts and calculate FPKM (fragments per kilobase of transcript per million mapped reads) and TPM (transcripts per million) values.

5. Quality Control of Quantification:

• Use exploratory data analysis techniques like PCA, hierarchical clustering, and negative log expression plots to assess sample similarity and identify outliers.
• Examine the proportion of reads aligning to the top 10 transcripts and the proportion of genes with zero read counts, which can highlight potential issues related to RNA fraction or library preparation.

6. Pre-processing for Differential Expression Analysis:

• Filter out genes with low read counts to focus on reliable measurements and minimize noise.
• Account for library size biases by normalizing read counts across samples.
• Address batch effects using methods like RUVseq, which aims to identify and remove latent factors that may introduce variability between batches.

7. Differential Expression Analysis:

• Use a statistical model that accounts for the discrete nature of read counts, like the negative binomial model.
• Tools like DESeq2 are widely used for differential expression analysis, incorporating shrinkage methods to improve the estimation of dispersion.
• Conduct multiple testing correction (e.g., using false discovery rate) to account for the large number of comparisons.

Conclusion

RNA-Seq has become an indispensable tool for studying gene expression, offering valuable insights into the connection between genetics and phenotype. By understanding the key steps involved in designing, executing, and analyzing RNA-Seq experiments, researchers can confidently leverage this technology to unravel the intricacies of biological pathways and mechanisms.