Transcriptomics

Fundamentals of Transcriptomics and RNA Biology

• Definition of Transcriptomics:

  • Transcriptomics is the study of the complete set of RNA molecules (transcripts) expressed by an organism, tissue, or cell type under specific developmental stages or physiological conditions.

• The Central Dogma of Molecular Biology:

  • Genetic information flows sequentially from deoxyribonucleic acid ($\text{DNA}$) to ribonucleic acid ($\text{RNA}$) via transcription, and subsequently from $\text{RNA}$ to protein via translation.

• Primary RNA Classifications:

  • Coding RNA:

  • Messenger RNA ($\text{mRNA}$): Linear single-stranded RNA transcripts that convey genetic code from nuclear $\text{DNA}$ to ribosomal translation machinery for polypeptide synthesis.

  • Non-coding RNA ($\text{ncRNA}$): Functional RNA molecules that do not translate into proteins but mediate core catalytic, infrastructural, and regulatory functions:

    • Ribosomal RNA ($\text{rRNA}$): Catalytic and structural components that coordinate with ribosomal proteins to assemble ribosomes and synthesise polypeptides.

    • Transfer RNA ($\text{tRNA}$): Adaptor molecules that decipher $\text{mRNA}$ codons and deliver corresponding amino acids to the nascent polypeptide chain during translation.

• Major Classes of Regulatory Non-Coding RNAs:

RNA Extraction, Isolation, and Small RNA Purification

• Total RNA Extraction via Guanidine Thiocyanate-Phenol (TRIzol):

  • Step 1: Cell Lysis and Sample Preparation:

  • Biological tissues or cell pellets are homogenized (thoroughly broken down and mixed) in a monophasic solution of guanidine thiocyanate and phenol (TRIzol / TRI Reagent) to disrupt cell membranes and dissolve intracellular components while inhibiting endogenous ribonucleases ($\text{RNases}$) (stopping the RNA-destroying enzymes that are naturally present inside a cell or tissue)

  • Step 2: Phase Separation Initiation:

  • Addition of chloroform induces phase separation following thorough mixing.

  • Step 3: Centrifugation and Phase Isolation:

  • Incubation and centrifugation separate the mixture into three phases: a lower red organic phase (containing proteins), an interphase (containing $\text{DNA}$), and an upper transparent aqueous phase containing total $\text{RNA}$.

  • The RNA-containing aqueous phase is carefully transferred into a clean tube.

  • Step 4: RNA Precipitation:

  • Pure isopropanol is added to the aqueous phase to precipitate total $\text{RNA}$.

  • Step 5: Pellet Sedimentation and Washing:

  • Centrifugation pellets the insoluble $\text{RNA}$ ; the isopropanol supernatant is decanted, and the pellet is air-dried.

  • Step 6: Pellet Resuspension:

  • The purified $\text{RNA}$ pellet is resuspended in nuclease-free water.

End-to-End Small RNA Purification and Sequencing Workflow ($18\text{--}30\,\text{nt}$):

• Step 1: Sample Collection and Total RNA Extraction:

• Material sources: Plant tissues (leaves, seeds), animal tissues, or in vitro cultured cells.

• Common kits/reagents: QIAGEN miRNeasy Mini/Micro Kit, Norgen Total RNA Purification Kit, TRIzol / TRI Reagent.

• Plant-specific procedural modifications: Polysaccharides and polyphenol compounds co-purify with nucleic acids and potently inhibit downstream enzymatic ligations. Inclusion of polyvinylpyrrolidone ($\text{PVP}$) and $\beta\text{-mercaptoethanol}$ during extraction is mandatory. Tissues from crops such as rice, wheat, oat, and tef require the miRNeasy Plant protocol or modified CTAB/TRIzol protocols. Plants characteristically exhibit abundant $24\,\text{nt}$ small interfering RNAs ($\text{siRNAs}$), elevating library complexity.

• Step 2: Quality Assessment of Extracted RNA:

• Fluorometric Concentration: Measured via Qubit fluorometer using the RNA HS (High Sensitivity) assay.

• Spectrophotometric Purity: Evaluated on a NanoDrop spectrophotometer; optimal pure RNA exhibits an $A_{260}/A_{280}$ ratio of approximately $2.0$.

• Electrophoretic Integrity: Profiled using an Agilent Bioanalyzer with RNA Nano or RNA Pico chips. Although the RNA Integrity Number ($\text{RIN}$) is less critical for small RNA sequencing than for intact $\text{mRNA}$ sequencing, high-integrity input RNA ensures robust library representation.

• Step 3: Small RNA Enrichment ($18\text{--}30\,\text{nt}$):

• Method A (Gold Standard): Gel-based size selection: Separation across a $15\%$ denaturing urea-polyacrylamide gel ($\text{PAGE}$) followed by targeted physical excision of the band corresponding to $18\text{--}30\,\text{nt}$.

• Method B: Commercial spin-column kits (Accelerated workflow): QIAGEN miRNA Isolation Kit, Zymo RNA Clean & Concentrator, NEB Monarch RNA Cleanup.

• Method C: Solid Phase Reversible Immobilization (SPRI) magnetic bead selection: Selective bead-to-sample volumetric ratios bind and remove high-molecular-weight RNAs ($\text{rRNA}$, $\text{mRNA}$), leaving small RNAs in suspension.

• Step 4: Yield of Purified Small RNA:

• Generates isolated fractions of enriched $18\text{--}30\,\text{nt}$ species ($\text{miRNAs}$, $\text{siRNAs}$, $\text{piRNAs}$, and $\text{tRNA}$-derived fragments) depleted of large $\text{rRNA}$ and $\text{mRNA}$.

• Step 5: Library Preparation:

• Standard biochemical cascade: Sequential $3'$ adapter ligation, $5'$ adapter ligation, reverse transcription to generate single-stranded $\text{cDNA}$, followed by high-fidelity $\text{PCR}$ amplification.

• Library size selection: Purifies the amplified product to isolate fragments of ~$140\text{--}160\,\text{bp}$ (accounting for the $18\text{--}30\,\text{nt}$ insert plus flanking adapter constructs).

• Commercial library kits: Illumina TruSeq Small RNA, QIAseq miRNA Library Kit, NEBNext Small RNA, NEXTFLEX Small RNA-seq Kit.

• Step 6: Library Quality Control:

• Validated on an Agilent Bioanalyzer using a High Sensitivity DNA Chip or an Agilent TapeStation D1000 screen.

• Confirms a narrow, distinct electrophoretic peak centered at ~$145\text{--}155\,\text{bp}$

• Step 7: High-Throughput Sequencing:

• Conducted on Illumina platforms (e.g., NextSeq, NovaSeq) using single-end sequencing protocols, typically with read lengths of $50\,\text{bp}$.

• Step 8: Downstream Computational Analysis:

• Successive pipeline: Raw read quality control and adapter trimming $\rightarrow$ Genome alignment against a reference sequence $\rightarrow$ Quantification of specific small RNA populations ($\text{miRNAs}$, $\text{siRNAs}$, $\text{piRNAs}$) $\rightarrow$ Differential expression statistical modeling $\rightarrow$ Functional annotation and target prediction.

Complementary DNA (cDNA) Synthesis

• Definition and Properties of cDNA:

  • Complementary $\text{DNA}$ ($\text{cDNA}$) is synthetic DNA synthesized (the process of creating or building new strands of deoxyribonucleic acid (DNA)) enzymatically from an $\text{RNA}$ template via the enzyme reverse transcriptase.

  • Because eukaryotic mature $\text{mRNA}$ (a fully processed molecule of messenger RNA that is ready to be translated into proteins by the cell's ribosome) undergoes post-transcriptional splicing, $\text{cDNA}$ derived from mature $\text{mRNA}$ contains exclusively coding exons without intervening non-coding introns.

• Biochemical Synthesis Mechanism:

  • Annealing: Mature eukaryotic $\text{mRNA}$ molecules carrying a $3'$ polyadenylated tail ($\text{poly(A)}$) are hybridized with a complementary synthetic oligo-dT primer ($\text{poly(T)}$).

  • First-Strand cDNA Synthesis: Reverse transcriptase utilizes the hybridized oligo-dT primer to polymerize a complementary antisense single-stranded $\text{DNA}$ strand moving in the $5'$ to $3'$ direction.

  • Second-Strand cDNA Synthesis: DNA polymerase synthesizes the complementary sense strand using the first-strand $\text{cDNA}$ as a template, producing a double-stranded $\text{cDNA}$ molecule suitable for cloning or sequencing.

Microarray Technology and Applications

• Historical Context and Operating Principles:

  • First published in 1995.

  • Microarrays quantify the relative abundance of a predetermined set of target transcripts based on nucleic acid hybridization.

  • Substrates consist of solid matrices (microscopic glass slides, specialized polymers, or micro-beads) onto which high-density grids of single-stranded synthetic oligonucleotide probes are covalently tethered.

  • Microarrays = a laboratory tool used to detect and measure the expression of thousands of genes or analyze genetic variations (like single-nucleotide polymorphisms or SNPs) all at the same time on a small solid surface, such as a glass slide or silicon chip

• Comparative Two-Channel Microarray Experimental Workflow:

  • Sample Preparation: Extraction of $\text{poly(A)+ mRNA}$ from distinct experimental groups (e.g., biological females versus males).

  • Labeling: $\text{mRNA}$ is reverse-transcribed into $\text{cDNA}$ using reverse transcriptase and random hexamer priming, simultaneously incorporating distinct fluorophores (e.g., green fluorophore for female-derived $\text{cDNA}$, red fluorophore for male-derived $\text{cDNA}$).

  • Hybridization: Labeled $\text{cDNA}$ populations are pooled and incubated on the microarray substrate to anneal competitively to complementary immobilized oligonucleotide probes.

  • Washing: Microarray substrates are washed under stringency conditions to eliminate unhybridized and non-specifically bound nucleic acids.

  • Laser Acquisition and Signal Processing: Dual-wavelength confocal laser scanning excites fluorophores at respective absorption maxima, generating image files (several megabytes in size) showing spot-specific fluorescent emission intensities.

  • Bioinformatic Classification: Processed text files compute differential intensity ratios, categorizing loci into female-biased, male-biased, or non-sex-biased expression profiles.

• Limitations of Microarray Architectures:

  • Microarrays are closed systems limited to interrogated transcripts; unannotated genes, novel isoforms, or sequences lacking complementary probes on the array cannot be quantified.

• Microarrays for Single Nucleotide Polymorphism (SNP) Interrogation:

  • Probe Design: Substrates are synthesized with immobilized allele-specific oligonucleotides terminating at or spanning targeted single-nucleotide variants ($\text{A}$, $\text{T}$, $\text{C}$, or $\text{G}$).

  • Genomic Breeding and Marker Pipelines:

  • In Silico Mining: Querying public repositories (dbSNP database) identifies prospective $\text{SNP}$ pools.

  • Primary Selection (Validation): High- to medium-throughput validation assays screen candidate germplasm pools against identified $\text{SNPs}$.

  • Secondary Selection (On-farm Selection): Medium- to low-throughput screening selects validated markers associated with specific Quantitative Trait Loci ($\text{QTL}$) on targeted chromosomes (e.g., chromosome 1 and chromosome 4 in swine models) for genomic selection and association mapping.

RNA Sequencing (RNA-Seq) Methodology and Comparison to Microarrays

• RNA-Seq Technology Overview:

  • First published in 2006.

  • High-throughput next-generation sequencing ($\text{NGS}$) enables direct base-by-base sequencing of $\text{cDNA}$ fragments generated from cellular transcriptomes.

  • Library Construction: Total or fractionated $\text{RNA}$ is isolated, sheared into fragments, converted to double-stranded $\text{cDNA}$, and end-repaired; sequencing adaptors are ligated to one or both termini.

  • Sequencing Strategy: High-throughput sequencing yields either single-end reads (sequenced from one terminus) or paired-end reads (sequenced inward from both termini), outputted into standard text-based FASTQ files.

  • Read Dimensions: Typically ranges from $30\,\text{bp}$ to $400\,\text{bp}$ on short-read platforms.

  • Signal Fidelity: Exhibits low optical and hybridization background noise compared to microarrays.

• Technological Benchmark: RNA-Seq vs. Microarrays (DOI: 10.1371/journal.pcbi.1005457):

• 
  

  

  Comprehensive Survey of High-Throughput RNA-Seq Platforms:

  • 454 (Roche):

  • Release year: 2005.

  • Typical read length: $700\,\text{bp}$.

  • Maximum throughput per run: $0.7\,\text{Gbp}$.

  • Single-read accuracy: $99.9\%$.

  • Primary application: Historical early-generation RNA-seq.

  • Illumina:

  • Release year: 2006.

  • Typical read length: $50\text{--}300\,\text{bp}$.

  • Maximum throughput per run: $>1000\,\text{Gbp}$.

  • Single-read accuracy: $>99.9\%$.

  • Primary applications: Differential gene expression ($\text{DGE}$), bulk transcriptome sequencing, single-cell RNA-seq ($\text{scRNA-seq}$).

  • Ion Torrent:

  • Release year: 2010.

  • Typical read length: $200\text{--}600\,\text{bp}$.

  • Maximum throughput per run: ~$30\,\text{Gbp}$.

  • Single-read accuracy: ~$98\%$.

  • Primary application: Targeted transcriptomics.

  • PacBio HiFi:

  • Release year: 2011.

  • Typical read length: $10\text{--}25\,\text{kb}$.

  • Maximum throughput per run: $20\text{--}100\,\text{Gbp}$.

  • Single-read accuracy: $>99.9\%$.

  • Primary applications: Full-length transcript sequencing without assembly, complex isoform profiling.

  • Oxford Nanopore:

  • Release year: 2014.

  • Typical read length: $1\,\text{kb}$ to $>100\,\text{kb}$.

  • Maximum throughput per run: Up to several $\text{Tb}$.

  • Single-read accuracy: $95\text{--}99\%+$.

  • Primary applications: Direct native RNA sequencing, comprehensive isoform identification, RNA modification mapping.

Sequencing Quality Control and Transcriptome Assembly

• Sequencing Quality Control Metrics and Filtering:

  • Phred Quality Scoring: Every called base is assigned an error probability score ranging from $0$ to $40$, where $Q = -10 \log_{10}(P_{\text{error}})$.

  • Quality Score Threshold: Base calls must exceed a Phred score threshold of $25$ to be retained without filtering.

  • FastQC Metric Interpretation:

  • Blue curve represents the mean quality score at each consecutive base pair window across the read length.

  • Central red line inside each yellow rectangular box denotes the median quality score.

  • Yellow box denotes the interquartile range ($\text{IQR}$, $25\text{th}$ to $75\text{th}$ percentile).

  • Whiskers demarcate the boundary of the $10\text{th}$ and $90\text{th}$ percentiles.

  • Green horizontal background indicates high quality ($Q > 28$); yellow background indicates acceptable warning zones ($20 \le Q \le 28$); red background indicates poor quality ($Q < 20$).

  • Trimming and Preprocessing Actions: Low-quality bases located at read termini (notably where $3'$ base qualities degrade below threshold, such as positions $285\text{--}300\,\text{bp}$), along with adapter sequences and experimental barcodes, must be trimmed prior to alignment.

• De Novo Transcriptome Assembly:

  • Reconstructs transcripts directly from overlapping sequence reads without aligning to a reference genome.

  • Core challenges include resolving complex contigs, eliminating chimeric assembly artifacts and sequencing errors, and optimizing algorithmic computational memory.

  • Assembly Quality Assessment Metrics:

  • Median contig length.

  • Total assembled contig count.

  • $N_{50}$ Metric Definition: Given an assembled contig dataset ordered by descending length, $N_{50}$ represents the sequence length of the shortest contig such that all contigs of equal or greater length collectively constitute at least $50\%$ of the total assembly length.

• Genome-Guided Transcriptome Assembly:

  • Aligns sequenced reads directly against an established reference genome, using mapped coordinates to reconstruct splicing isoforms.

  • Follows alignment strategies similar to whole-genome sequencing workflows.

  • Quality is benchmarked by evaluating de novo structural integrity parameters and verifying alignment congruency against known, validated transcript annotations.

Gene Expression Quantification and Mathematical Normalization

• Read Count Quantification Principles:

  • Quantifies transcriptional variations triggered by experimental conditions, genetic interventions, or disease states.

  • Computes the aggregate tally of reads aligned to each specific genetic locus.

  • Raw read counts cannot be directly compared across different genes or experiments without mathematical normalization for sequencing depth and gene length.

• Library Size and Sequencing Depth Normalization:

  • Fragments Per Million Mapped Reads ($\text{FPM}$): Corrects fragment abundance based on total mapped library size in millions.

  • Reads Per Million Mapped Reads ($\text{RPM}$) / Counts Per Million ($\text{CPM}$): Normalizes raw counts against the total number of reads sequenced per sample.

• Gene Length Normalization:

  • Longer genes generate a proportionally higher number of fragmented sequencing reads than shorter genes expressed at identical molar concentrations.

  • Dividing depth-normalized values by feature length (in kilobases) generates normalized expression values.

• Formulas for Expression Normalization:

  • Reads Per Kilobase Million ($\text{RPKM}$): Applied exclusively to single-end RNA-seq datasets:

$RPKM = \frac{\text{ExonMappedReads} \times 10^9}{\text{TotalMappedReads} \times \text{ExonLength}}$

• Fragments Per Kilobase Million ($\text{FPKM}$): Applied to paired-end RNA-seq datasets, where two paired reads originate from a single continuous $\text{cDNA}$ fragment, preventing double-counting:

$FPKM = \frac{\text{ExonMappedFragments} \times 10^9}{\text{TotalMappedFragments} \times \text{ExonLength}}$

Bioinformatics Pipelines for Differential Gene Expression (DGE)

• The Tuxedo Suite / Galaxy Pipeline Architecture:

  • Bowtie: Memory-efficient, ultra-fast short-read aligner based on the Burrows-Wheeler Transform; indexes the reference genome (.fa file) and aligns short reads.

  • TopHat: Splicing-aware aligner that utilizes Bowtie alignments to identify exon-exon splice junctions and map reads spanning introns across the genome.

  • Cufflinks: Assembles aligned reads into individual transcripts, constructs parsimonious isoform models, and computes relative expression abundance based on gene annotations (.gtf file).

  • Cuffmerge: Merges assembled transcript structures across multiple experimental conditions and replicates into a unified transcriptome assembly.

  • Cuffdiff: Calculates differential expression between comparative experimental conditions (e.g., condition $A$ versus condition $B$), testing significance through statistical models of transcript abundance.

  • CummeRbund: Downstream R-based computational environment designed to parse, index, and visualize Cuffdiff statistical output files.

• Modern Count-Based Differential Gene Expression Pipeline:

  • Sequence Ingestion: Input raw paired-end or single-end reads in FASTQ format.

  • Splice-Aware Sequence Alignment: Reads are mapped against the reference genome using modern high-efficiency aligners (HISAT2, STAR, TopHat2, or Bowtie 2).

  • Alignment Compression: Alignment coordinates are written to a Binary Alignment Map ($\text{BAM}$) file, the compressed binary equivalent of a Sequence Alignment/Map ($\text{SAM}$) file.

  • Read Assignment and Quantification: Dedicated summarization tools (featureCounts or HTSeq) map reads in $\text{BAM}$ files against genomic features, generating a matrix of integer read counts (gene identifiers versus experimental conditions/replicates).

  • Statistical Testing via DESeq2:

  • Uses negative binomial generalized linear models to analyze the read count matrix.

  • Evaluates pair-wise sample groups to calculate normalized fold changes ($\log_2(\text{fold change})$, indicating up- or down-regulation) and Benjamini-Hochberg false discovery rate-adjusted p-values ($p_{\text{adj}}$) to establish statistical significance for Differential Gene Expression ($\text{DGE}$).