# Lecture 17: Next-Generation Sequencing and Genomics

## Genetics

---

## Learning Objectives

By the end of this lecture, students will be able to:

1. Explain the principles of next-generation sequencing (NGS) technologies
2. Compare Illumina sequencing, Ion Torrent, PacBio, and Oxford Nanopore platforms
3. Describe the workflow of whole-genome sequencing and whole-exome sequencing
4. Explain bioinformatic approaches to genome assembly and variant calling
5. Discuss the Human Genome Project and its impact
6. Describe applications of genomics in research and clinical medicine

---

## Lecture Content

### I. From Sanger to Next-Generation Sequencing

Sanger sequencing, while highly accurate, is limited in throughput (approximately 700-1,000 base pairs per run) and expensive for large-scale projects. The Human Genome Project (1990-2003) cost approximately $2.7 billion using Sanger sequencing. **Next-generation sequencing (NGS)** technologies revolutionized the field through massively parallel sequencing, generating millions to billions of reads simultaneously and dramatically reducing cost. A whole human genome can now be sequenced for approximately $200-600.

Several key concepts underpin all NGS platforms. **Library preparation** involves fragmenting DNA and ligating adapters to the fragment ends. **Clonal amplification** generates multiple copies of each fragment (as clusters or on beads) to improve the detection signal. **Sequencing by synthesis (SBS)** adds nucleotides one at a time, detecting each incorporation event. **Coverage (depth)** refers to the average number of times each base is sequenced; for example, 30x coverage means each position is read an average of 30 times.

### II. Illumina Sequencing (Sequencing by Synthesis)

Illumina is the most widely used NGS platform. The workflow begins with **library preparation**, in which DNA is fragmented to approximately 200-500 base pairs, end-repaired, A-tailed, and ligated to adapters. During **cluster generation**, single-stranded library molecules bind to complementary oligonucleotides on the flow cell surface and undergo bridge amplification: the DNA bends over, anneals to an adjacent oligonucleotide, and is extended, producing clonal clusters of approximately 1,000 copies each.

**Sequencing by synthesis** then proceeds using fluorescently labeled reversible terminators, one for each base. In each cycle, a single nucleotide is incorporated, fluorescence is detected through imaging, and then the terminator and fluorophore are chemically removed before the next cycle. Read lengths range from 75 to 300 base pairs per end. **Paired-end sequencing**, which sequences both ends of each fragment, improves alignment accuracy and enables detection of structural variants. Output ranges from approximately 25 million reads on the MiSeq to billions of reads on the HiSeq/NovaSeq platforms, with an error rate of approximately 0.1% per base (dominated by substitution errors). The advantages of Illumina sequencing are high accuracy, high throughput, and low cost per base. Its limitations include short read length, GC bias, and difficulty with repetitive regions.

### III. Long-Read Sequencing Technologies

**Pacific Biosciences (PacBio) SMRT sequencing** (Single Molecule Real-Time) fixes a DNA polymerase at the bottom of a zero-mode waveguide (ZMW) and detects fluorescently labeled nucleotides as they are incorporated in real time. Read lengths average 10-25 kb with some exceeding 100 kb. While individual reads have a higher error rate (approximately 10-15%, with random errors), circular consensus sequencing (HiFi reads) achieves accuracy above 99.9%. PacBio also detects DNA modifications such as methylation directly.

**Oxford Nanopore sequencing** threads single-stranded DNA or RNA through a protein nanopore embedded in a membrane, detecting changes in ionic current as each base passes through to determine base identity. Read lengths are theoretically unlimited, with reads exceeding 4 Mb achieved and typical averages of 10-100 kb. The technology enables real-time sequencing on portable devices (the MinION is USB-sized). The error rate is currently approximately 5-10% but is improving. Nanopore sequencing can directly sequence native DNA and RNA, detecting modifications without chemical pretreatment. Applications include field diagnostics, rapid pathogen identification, and real-time outbreak surveillance.

### IV. The Human Genome Project and Key Findings

The Human Genome Project was an international effort spanning 1990-2003, involving both a public consortium (International Human Genome Sequencing Consortium) and a private effort (Celera Genomics, led by Craig Venter), with the draft genome published in 2001. Key findings revealed that the human genome contains approximately 3.2 billion base pairs and roughly 20,000-25,000 protein-coding genes, far fewer than the 100,000 initially expected (the "gene number paradox"). Protein-coding sequences represent only about 1.5% of the genome, while over 50% consists of repetitive sequences including transposable elements and satellite DNA. Any two humans share approximately 99.9% sequence identity. Gene density varies by chromosome, with chromosome 19 being gene-dense and chromosome 13 being gene-poor.

The subsequent **ENCODE project** (Encyclopedia of DNA Elements) aimed to identify all functional elements in the human genome and found that approximately 80% of the genome has some biochemical activity, being transcribed, bound by proteins, or carrying chromatin modifications. The functional significance of this widespread biochemical activity versus biological noise remains a subject of debate.

<image>Panel A: Illumina sequencing workflow diagram showing the four main steps: library preparation (fragmentation, adapter ligation), bridge amplification on the flow cell surface creating clonal clusters, sequencing by synthesis with fluorescent reversible terminators (showing one cycle of incorporation, imaging, and cleavage), and data output as a FASTQ file. Panel B: Comparison diagram of short-read (Illumina) vs. long-read (PacBio/Nanopore) sequencing: short reads shown as many small fragments aligned to a reference with difficulty spanning repetitive regions, vs. long reads spanning entire repeats and structural variants. Panel C: Pie chart of human genome composition: protein-coding exons (1.5%), introns (25%), intergenic DNA (25%), transposable elements (45%), and other (3.5%), with a separate bar showing the ~20,000 protein-coding genes.</image>

### V. NGS Applications and Workflows

**Whole-Genome Sequencing (WGS)** sequences the entire genome and can detect SNVs, indels, structural variants, and copy number variants. Clinical applications typically require approximately 30x coverage. WGS offers the advantage of being unbiased, capturing both coding and non-coding variants, but generates large data volumes with many variants of uncertain significance (VUS).

**Whole-Exome Sequencing (WES)** captures and sequences only the exonic regions, approximately 1-2% of the genome encompassing roughly 22,000 genes. Hybridization capture probes (bait sequences) enrich for exons, and approximately 100x coverage is required. WES is more cost-effective than WGS and identifies most disease-causing coding variants, but it misses non-coding regulatory variants, structural variants, and deep intronic mutations.

**Targeted panel sequencing** sequences a predefined set of genes (such as cancer gene panels or cardiac gene panels) at high coverage (over 500x), providing high sensitivity for detecting low-frequency variants like somatic mutations in tumors. **RNA-Seq** sequences all RNA transcripts in a sample, quantifying gene expression levels, identifying alternative splicing events, and detecting fusion transcripts; it has largely replaced microarrays. **ChIP-Seq** identifies genome-wide protein binding sites for transcription factors and histone modifications. **ATAC-Seq** maps open chromatin regions. **Metagenomics** sequences all DNA in an environmental or clinical sample to identify microbial communities.

### VI. Bioinformatics Pipeline for NGS Data

The standard bioinformatics pipeline begins with **quality control**, assessing raw read quality using tools like FastQC and trimming adapters and low-quality bases with Trimmomatic. **Alignment** maps reads to a reference genome using BWA or Bowtie2 for short reads, or minimap2 for long reads. **Variant calling** identifies differences from the reference using tools such as GATK HaplotypeCaller or DeepVariant for SNVs and indels, and Manta or DELLY for structural variants. **Annotation** determines the functional impact of each variant using databases and tools including ANNOVAR, VEP/Ensembl, ClinVar, and gnomAD. **Filtering and interpretation** applies criteria such as allele frequency, predicted pathogenicity scores (CADD, SIFT, PolyPhen-2), and clinical database entries to prioritize clinically relevant variants.

For **de novo genome assembly** when no reference genome exists, algorithms based on overlap-layout-consensus or de Bruijn graphs are employed. Standard data formats include FASTQ for raw reads, BAM/SAM for aligned reads, and VCF for variant calls.

<image>Panel A: Bioinformatics pipeline flowchart for NGS data analysis: raw FASTQ reads → quality control → adapter trimming → alignment to reference genome (BAM file) → variant calling (VCF file) → annotation (functional impact) → filtering and clinical interpretation, with software tools labeled at each step. Panel B: Diagram comparing whole-genome sequencing vs. whole-exome sequencing: WGS shows reads covering the entire genome uniformly, while WES shows reads concentrated on exonic regions after hybridization capture, with an illustration of the capture probe hybridization step. Panel C: Coverage plot showing read depth across a genomic region, with peaks over exons in WES data, and explanation of coverage metrics (e.g., 30x mean coverage, with areas of low coverage potentially missing variants).</image>

---