7.2 RNA-Seq

RNA-Seq revolutionizes how we study gene expression. It allows us to see the full picture of RNA in cells, giving insights into which genes are active and how they're spliced. This powerful tool has become essential for understanding how genes work in different conditions.

The RNA-Seq process involves careful sample prep, turning RNA into DNA libraries, and sequencing. After sequencing, the data goes through quality control, alignment to a reference, and analysis to find important differences in gene activity between samples.

Overview of RNA-Seq

• Powerful high-throughput sequencing technique revolutionizes transcriptome analysis in bioinformatics
• Enables comprehensive profiling of RNA molecules present in a biological sample at a given time
• Provides insights into gene expression patterns, alternative splicing events, and novel transcript discovery

RNA-Seq workflow

Sample preparation

Top images from around the web for Sample preparation

• 

  Frontiers | Single-Cell RNA-Seq Analysis Uncovers Distinct Functional Human NKT Cell Sub ... View original

  Is this image relevant?

• 

  Frontiers | A Scalable Strand-Specific Protocol Enabling Full-Length Total RNA Sequencing From ... View original

  Is this image relevant?

• 

  Frontiers | Direct Metatranscriptome RNA-seq and Multiplex RT-PCR Amplicon Sequencing on ... View original

  Is this image relevant?

• 

  Frontiers | Single-Cell RNA-Seq Analysis Uncovers Distinct Functional Human NKT Cell Sub ... View original

  Is this image relevant?

• 

  Frontiers | A Scalable Strand-Specific Protocol Enabling Full-Length Total RNA Sequencing From ... View original

  Is this image relevant?

1 of 3

Top images from around the web for Sample preparation

• 

  Frontiers | Single-Cell RNA-Seq Analysis Uncovers Distinct Functional Human NKT Cell Sub ... View original

  Is this image relevant?

• 

  Frontiers | A Scalable Strand-Specific Protocol Enabling Full-Length Total RNA Sequencing From ... View original

  Is this image relevant?

• 

  Frontiers | Direct Metatranscriptome RNA-seq and Multiplex RT-PCR Amplicon Sequencing on ... View original

  Is this image relevant?

• 

  Frontiers | Single-Cell RNA-Seq Analysis Uncovers Distinct Functional Human NKT Cell Sub ... View original

  Is this image relevant?

• 

  Frontiers | A Scalable Strand-Specific Protocol Enabling Full-Length Total RNA Sequencing From ... View original

  Is this image relevant?

1 of 3

• Involves careful extraction of RNA from biological samples (cells, tissues, organisms)
• Requires preservation of RNA integrity using specialized reagents and protocols
• Includes removal of genomic DNA contamination through DNase treatment
• Often incorporates ribosomal RNA depletion or poly-A selection to enrich for mRNA

Library construction

• Converts RNA into cDNA through reverse transcription
• Fragments cDNA to desired sizes (typically 200-500 bp)
• Adds sequencing adapters to cDNA fragments through ligation or PCR amplification
• Incorporates unique molecular identifiers (UMIs) to reduce PCR bias and improve quantification accuracy

Sequencing platforms

• Illumina sequencing dominates RNA-Seq applications due to high throughput and low error rates
• Ion Torrent offers faster sequencing runs with comparable accuracy to Illumina
• Pacific Biosciences and Oxford Nanopore provide long-read sequencing capabilities for improved isoform detection
• Sequencing depth varies depending on experimental goals (typically 10-30 million reads per sample)

Read quality control

Quality metrics

• Phred quality scores assess base-calling accuracy on a logarithmic scale
• Per-base sequence quality evaluates the average quality score at each position across all reads
• GC content distribution helps identify potential contamination or bias
• Sequence duplication levels indicate PCR amplification bias or low library complexity

Trimming and filtering

• Removes low-quality bases from read ends to improve alignment accuracy
• Trims adapter sequences that may interfere with downstream analysis
• Filters out reads with overall low quality or high proportion of ambiguous bases (N's)
• Discards reads shorter than a specified threshold after trimming to maintain mapping efficiency

Read alignment

Reference genome vs transcriptome

• Genome alignment maps reads to the entire genomic sequence
  • Allows detection of novel splice junctions and unannotated transcripts
  • Requires more computational resources and time
• Transcriptome alignment maps reads to known transcript sequences
  • Faster and less computationally intensive
  • Limited to detecting expression of known genes and isoforms

Alignment algorithms

• Burrows-Wheeler Transform (BWT) based aligners (BWA, Bowtie2) offer fast and memory-efficient alignment
• Hash table-based aligners (STAR, HISAT2) provide improved speed for large genomes
• Suffix array-based aligners (Subread, SOAP2) balance speed and sensitivity
• Each algorithm employs different strategies for handling mismatches and gaps

Splice-aware mapping

• Identifies and aligns reads spanning exon-exon junctions
• Utilizes known splice site information from gene annotations
• Employs specialized algorithms to detect novel splice junctions
• Improves accuracy of transcript quantification and alternative splicing analysis

Transcript assembly

De novo assembly

• Reconstructs transcripts without a reference genome
• Utilizes graph-based algorithms (de Bruijn graphs) to assemble reads into contigs
• Effective for non-model organisms or discovering novel transcripts
• Computationally intensive and sensitive to sequencing errors and coverage depth

Reference-guided assembly

• Combines reference genome information with RNA-Seq data to improve transcript reconstruction
• Identifies novel isoforms and refines existing gene models
• Reduces computational complexity compared to de novo assembly
• Allows for the detection of fusion transcripts and gene rearrangements

Quantification of gene expression

Read counting methods

• Simple counting assigns reads to genes based on overlap with exonic regions
• Unique molecular identifier (UMI) counting reduces PCR amplification bias
• Transcript-level quantification estimates abundance of individual isoforms
• Pseudoalignment methods (Kallisto, Salmon) provide rapid quantification without full alignment

Normalization techniques

• RPKM (Reads Per Kilobase Million) normalizes for sequencing depth and gene length
• TPM (Transcripts Per Million) improves upon RPKM by ensuring the sum of normalized values is constant
• DESeq2's median of ratios method accounts for differences in library size and RNA composition
• TMM (Trimmed Mean of M-values) normalizes based on the assumption that most genes are not differentially expressed

Differential expression analysis

Statistical models

• Negative binomial distribution models count data variability in RNA-Seq experiments
• Generalized linear models (GLMs) account for complex experimental designs and covariates
• Empirical Bayes methods shrink gene-wise dispersion estimates to improve stability
• Likelihood ratio tests or Wald tests assess significance of differential expression

Multiple testing correction

• Controls for false positives when performing thousands of statistical tests simultaneously
• Benjamini-Hochberg procedure controls the false discovery rate (FDR)
• Bonferroni correction provides stringent control of the family-wise error rate (FWER)
• q-values represent the minimum FDR at which a test may be called significant

Functional annotation

Gene ontology analysis

• Categorizes differentially expressed genes into functional groups
• Utilizes standardized vocabulary to describe gene functions, processes, and cellular components
• Performs enrichment analysis to identify overrepresented GO terms in gene sets
• Considers the hierarchical structure of GO terms to avoid redundancy in results

Pathway enrichment

• Identifies biological pathways significantly affected by differentially expressed genes
• Utilizes databases like KEGG, Reactome, or BioCarta for pathway information
• Applies statistical methods (Fisher's exact test, GSEA) to assess pathway enrichment
• Visualizes pathway interactions and gene expression changes in network diagrams

Visualization of RNA-Seq data

Heatmaps

• Represent gene expression levels across samples using color intensity
• Allow for hierarchical clustering of genes and samples to reveal patterns
• Incorporate dendrograms to show relationships between clusters
• Can be customized with additional annotations (sample groups, gene functions)

Volcano plots

• Display statistical significance vs fold change for all genes in a differential expression analysis
• X-axis represents log2 fold change, Y-axis represents -log10 adjusted p-value
• Helps identify genes with both large magnitude changes and statistical significance
• Can be enhanced with gene labels, color coding, or interactive features

Principal component analysis

• Reduces high-dimensional gene expression data to a few principal components
• Visualizes sample relationships in a low-dimensional space (typically 2D or 3D)
• Reveals major sources of variation in the dataset
• Helps identify batch effects or unexpected sample clustering

Challenges in RNA-Seq analysis

Batch effects

• Arise from technical variations between sample preparation or sequencing runs
• Can confound biological signals and lead to false discoveries
• Addressed through experimental design (sample randomization) and computational methods (ComBat, RUVSeq)
• Requires careful consideration in meta-analyses combining multiple datasets

Low-abundance transcripts

• Difficult to accurately quantify due to limited sequencing depth
• May require increased sequencing depth or targeted approaches for detection
• Affected by competition from highly expressed genes during library preparation
• Can be partially mitigated through normalization techniques and statistical modeling

Alternative splicing detection

• Challenges in distinguishing true isoforms from sequencing or alignment artifacts
• Requires sufficient read coverage across splice junctions for accurate detection
• Complicated by the presence of novel or unannotated splice variants
• Necessitates specialized algorithms and statistical models for robust identification

Applications of RNA-Seq

Gene expression profiling

• Measures abundance of transcripts across different conditions or time points
• Enables identification of differentially expressed genes in disease states
• Facilitates discovery of biomarkers for diagnosis or prognosis
• Provides insights into regulatory networks and cellular processes

Novel transcript discovery

• Identifies previously unannotated genes or isoforms
• Reveals non-coding RNAs (lncRNAs, miRNAs) with potential regulatory functions
• Improves genome annotations and understanding of transcriptional complexity
• Contributes to the discovery of species-specific or tissue-specific transcripts

Fusion gene detection

• Identifies chimeric transcripts resulting from chromosomal rearrangements
• Important for cancer research and diagnosis of certain genetic disorders
• Requires specialized algorithms to detect reads spanning fusion breakpoints
• Can be validated through targeted approaches (RT-PCR, FISH)

RNA-Seq vs microarrays

Sensitivity and specificity

• RNA-Seq offers higher sensitivity for detecting low-abundance transcripts
• Provides better specificity in distinguishing between similar sequences
• Allows for detection of novel transcripts and splice variants
• Microarrays limited by probe design and cross-hybridization issues

Dynamic range comparison

• RNA-Seq exhibits a wider dynamic range (>10^5) for accurate quantification
• Microarrays suffer from signal saturation for highly expressed genes
• RNA-Seq provides more accurate fold-change estimates for differential expression
• Enables detection of subtle expression changes that may be missed by microarrays

Emerging RNA-Seq technologies

Single-cell RNA-Seq

• Profiles transcriptomes of individual cells rather than bulk populations
• Reveals cellular heterogeneity and rare cell types within complex tissues
• Utilizes specialized protocols for cell isolation and library preparation (Drop-seq, 10x Genomics)
• Requires advanced computational methods to handle technical noise and sparsity

Long-read sequencing

• Produces reads spanning entire transcripts (up to several kilobases)
• Improves detection and quantification of alternative splicing events
• Facilitates de novo transcriptome assembly and isoform discovery
• Challenges include higher error rates and lower throughput compared to short-read sequencing