Illumina Sequencing by Synthesis Workflow Overview

Third-generation sequencing enables single-molecule sequencing without PCR amplification, allowing much longer read lengths (up to 100,000+ bp). Oxford Nanopore technology uses protein nanopores embedded in membranes; DNA molecules are threaded through nanopores one at a time, and as each base passes, it causes characteristic electrical current changes. The MinION device is portable and connects to laptops. Applications include genomics (whole genome and exome sequencing), epigenomics (DNA methylation studies), transcriptomics (RNA-seq), metagenomics (environmental sample analysis), and resequencing (comparing to reference genomes). ChIP-Seq studies protein-DNA interactions by immunoprecipitating specific proteins and sequencing bound DNA.

Sequencing applications include genomics (whole genome), exomics (exon-focused), transcriptomics (RNA), epigenomics (regulatory elements), and metagenomics (environmental DNA). ChIP-Seq maps protein-DNA interactions by sequencing DNA bound to specific proteins. Targeted sequencing (amplicon sequencing) focuses on specific genes, widely used for microbial community analysis via 16S rRNA gene sequencing. Single-cell sequencing analyzes individual cells to reveal cellular heterogeneity. Sequencing technologies evolved from Sanger (1970s) through second-generation (Illumina, 2000s) to third-generation (PacBio, Nanopore, 2010s). Illumina dominates due to high throughput, accuracy, and low cost per base. PacBio offers long reads with high accuracy but higher cost. Nanopore provides portability and real-time sequencing with improving accuracy.

NGS enables diverse applications: variant calling identifies genetic differences using VCF format, bisulfite sequencing maps DNA methylation, RNA-seq quantifies gene expression, ChIP-seq studies protein-DNA interactions, and ATAC-seq reveals chromatin accessibility. Metagenomics bypasses culture limitations by directly sequencing environmental DNA, enabling study of microbial communities through 16S rRNA sequencing (genus-level) or whole-genome shotgun sequencing (species-level). Major projects include the Human Microbiome Project and Earth Microbiome Project.

Next-generation sequencing revolutionized genomics by reducing costs from $100,000 to ~$1,000 per genome. Major platforms include Illumina (~80% market share), SOLiD, PacBio, and Ion Torrent, generating ~150-nucleotide reads. Different applications require different approaches: whole genome sequencing for comparative genomics, RNA-seq for transcriptome characterization, ChIP-seq for protein-DNA binding, and methyl-seq for epigenetic studies. Each application requires specific library preparations and computational pipelines to extract meaningful biological insights from massive datasets.

NGS enables whole genome sequencing, targeted genome sequencing, and epigenetic modification detection. Whole genome sequencing sequences the entire genome, while targeted sequencing focuses on specific regions. Epigenetic modifications like DNA methylation can be detected by sequencing modified DNA. Transcriptome sequencing (RNA-seq) sequences all RNA molecules in a cell, revealing gene expression patterns. RNA-protein interaction sequencing (CLIP-seq) identifies which RNAs bind to specific proteins. These applications have revolutionized research in genomics, transcriptomics, and metagenomics.

The Illumina Sequencing Analysis Viewer (SAV) is a viewing application that displays important quality metrics from sequencing runs without performing data analysis; users can load run data using the interop folder, run info file, and run parameters file, then analyze key metrics including Q30 scores (which indicate base-calling accuracy with values above Q30 representing greater than 99.9% accuracy), yield (data quantity passing filter), and various cycle-by-cycle metrics such as intensity, full width half max (FWHM) for focus assessment, percent base distribution for library diversity evaluation, and phasing/pre-phasing values to troubleshoot sequencing quality issues across multiple tabs including Analysis, Imaging, Summary, and Indexing.

After running bcl2fastq, a summary report is generated and saved in the run folder. This report provides: (1) Total yield and estimated run output; (2) Summary by lane including cluster density (target is ~885K per mm² for HiSeq); (3) Percentage of clusters passing filter (should be >80%, ideally >85%); (4) Phasing and pre-phasing values (should be below 0.5, indicating flow or reagent issues if elevated); (5) Percentage of Q30 bases (should be around 80% for good runs); (6) Mean quality score; and (7) Top honeycomb barcodes showing barcodes identified in the run but not specified in the sample sheet, helping identify undetermined targets.

Advanced diagnostic techniques for pattern flow cells include plotting percent occupied (wells with clusters) against percent passing filter—optimally clustered shows high values, underloaded shows positive slope, overloaded shows vertical slope. For NextSeq 1000/2000, four key metrics diagnose suboptimal clustering: percent passing filter, percent Q30, percent phi-x aligned, and percent loading concentration. Over-clustered runs show low passing filter, low Q30, lower phi-x alignment than spiked-in percentage, and loading concentration above 95%. Under-clustered runs show low passing filter, high Q30, higher phi-x alignment than expected, and loading concentration below 95%. Causes include loading concentration errors, denaturation issues, or library quality problems. Under-clustering affects yield minimally but over-clustering risks complete run failure.

Phasing and prefasing are technical phenomena in Illumina sequencing that cause quality score degradation toward the end of reads. Phasing occurs when DNA molecules in a cluster go out of sync during synthesis—some nucleotides remain attached, causing molecules to read the same base twice and lag behind. Prefasing occurs when defective labeled nucleotides cause premature addition of multiple bases, making some molecules ahead of the rest. Both create mixed signals, reducing signal-to-noise ratio and lowering quality scores. This explains why quality scores decrease toward the end of Illumina reads.

Phasing occurs when most molecules in a cluster show up in the correct channel but a small number show up in the channel representing the last cycle's base. For example, a cluster called as G can faintly appear in the C or A channel depending on the previous base. Pipeline 1.6 generates Gerald summary files showing phasing and pre-phasing values calculated for and applied to each lane. Previous pipeline versions showed applied values (average across all lanes), with calculated values stored in phasing XML files named 'sore_starting_cycle_number_uncore_phasing.xml'. Individual tile values are scored in text files with the same naming convention. Plotting phasing and pre-phasing by tile and/or by lane helps diagnose problems with the Peltier block or reagent flow.

This section covers the technical foundations of Illumina sequencing and competing long-read approaches. Illumina employs reversible fluorescent terminators that incorporate, image, and then chemically remove to enable continued synthesis, repeating for 100-300 cycles. Read length is limited to ~300 bases because enzymatic imperfections cause strands to diverge over time. Chemistry evolved from four-color (requiring complex compensation) to two-color (reducing costs) to single-color (using sequential imaging). Competing long-read technologies include Oxford Nanopore, which uses protein nanopores (1.8 nm) where single-stranded DNA passing through alters electrical current, enabling direct RNA sequencing and 2-million-base reads with 10-15% error rates. PacBio uses modified nucleotides with fluorescent dyes on phosphate groups, allowing real-time detection in 100-nm reaction chambers, generating 1-200 kilobase reads with exceptional consensus accuracy through circular consensus sequencing.

Sequencing technology has evolved through three major generations. First-generation Sanger sequencing (developed in the 1970s) was very expensive and slow, limiting its use to early genome projects. Second-generation short-read technologies like Illumina produce reads of 100-300 bases with high accuracy (~99.99%), making them cost-effective for large-scale projects. Third-generation long-read technologies from Oxford Nanopore and Pacific Biosciences produce much longer reads (up to 1 million bases for Nanopore, 15-20,000 bases for HiFi reads) but with lower accuracy (~90-95%). Modern approaches combine HiFi reads for accuracy with Nanopore ultra-long reads for bridging repetitive regions.

Pacific Biosciences SMRT (Single Molecule Real-Time) sequencing is a third-generation sequencing technology that enables real-time observation of DNA synthesis by DNA polymerase, producing exceptionally long reads of 20-60 kilobase pairs (compared to Illumina's 250 bp average) with the ability to detect base modifications like methylation, though it has higher error rates (11-15%) and costs ($700,000+ instrument) than Illumina; it is particularly valuable for de novo genome assembly, identifying novel gene isoforms, detecting disease-associated mutations, and studying epigenetic modifications.

Illumina dominates the sequencing market with 70-80% market share through its short-read sequencing technology, which offers significant advantages in cost (reduced from $3 million to under $1,000), speed, and accuracy (lower error rate) compared to Pacific Bioscience's long-read technology, though long-read sequencing provides more information per read and is better suited for larger genomes like humans; the speaker plans to sell Illumina stock if the company does not expand into the long-read segment within a few years, as this represents a key competitive risk despite Illumina's current strong market position.

PacBio (SMRT) sequencing represents long-read technology using polymerases tethered to wells in a picotiter plate. As nucleotides are incorporated, fluorescent signals are detected in real time at polymerase speed. This technology can sequence very long DNA molecules (tens of kilobases) in single reads, making it valuable for genome assembly when long contiguous sequences are needed. Oxford Nanopore sequencing passes single-stranded DNA through a protein nanopore, measuring electrical property changes unique to different sequences. This technology produces extremely long reads (hundreds of kilobases) in a compact device but currently has lower accuracy than Illumina or PacBio. Both technologies complement Illumina's short-read approach for comprehensive genomic analysis.