Genomics Notes

Genomics

• Genomics is the study of complete genomes, particularly the analysis of the complete DNA sequence of an organism's genome.
• The field of genomics became viable after the publication of the Haemophilus influenzae genome sequence in 1995.

Physical Mapping

• A physical map is a set of cloned DNA fragments with known positions relative to each other in the genome.
• The complete DNA sequence of a gene or genome is the ultimate physical map.
• Historically, intermediate-level physical maps were constructed from cloned fragments for sequencing and other manipulations.
• A large genome (e.g., a bacterial genome of $3 \, \text{Mb}$) can be subcloned into lambda phage vectors (carrying $15 - 20 \, \text{kb}$).
• The minimum number of clones required to cover the genome would be 2000, assuming no overlap.
• Overlapping clones are aligned and positioned on the chromosome relative to each other.
• Sets of clones, called contigs (contiguated clones), are checked for stability and representation of the starting genome.
• Large-insert vectors are commonly used for physical mapping:
  • Lambda phage: up to $20 \, \text{kb}$
  • Cosmid: < $35 \, \text{kb}$
  • Bacterial Artificial Chromosome (BAC): < $150 \, \text{kb}$
• Plasmid vectors may be viable for smaller sequencing projects.
• Clones are ordered by:
  • Hybridization: Labelled probes detect clones sharing sequences.
  • Fingerprinting
  • End-sequencing
• Hybridization methods use labeled probes to detect clones that share sequence.
• Probes can be generated from each end of the clone by "end rescue," or DNA fragments isolated can be used (e.g., cDNA clones).
• A problem with hybridization is that significant repeat content in the genome can cause some probes/clone ends to fail to provide a unique link to the next segment of the genome.
• In hybridization mapping, clones are picked into a grid, hybridized to probes, and contigs are built.
• Clones hybridizing to multiple probes are predicted to overlap, while those hybridizing to only one probe are predicted to extend outwards.
• The process is repeated until all clones are ordered with respect to each other and the chromosome.

Genome Sequencing

• Small genomes: 'Shotgun' (random) clone complete genome in fragments of < $1 \, \text{kb}$
• Medium/large genomes: Shotgun clone fragments of < $1 \, \text{kb}$ from ordered cosmid or BAC library
• Automated sequencing of overlapping shotgun libraries.
• Next-Gen Sequencing (NGS)
• Sequence assembly: Automated using sequence assembly software.
• Exponential increase in sequenced genomes due to Next-Gen Sequencing technologies.

Next-Generation Sequencing

• Steps include:
  • Template DNA verification (quality and quantity)
  • Fragmentation (hydrodynamic shearing, enzymatic methods)
  • End repair
  • Size selection
  • Adaptor ligation
  • Amplification (emulsion PCR, solid-phase bridge amplification)
  • Sequencing (various platforms like 454 GS Junior, Ion PGM, MiSeq, HiSeq, PacBio RS)

Next-Generation Sequencing Machines

• Overview of various sequencing machines, their chemistries, read lengths, run times, approximate costs, advantages, and disadvantages:
  • 454 GS FLX+ (Roche): Pyrosequencing, $700 - 800$ bases, 23 hours, $0.7 \, \text{Gb}$, long read lengths, high reagent costs, high error rate in homopolymers.
  • HiSeq 2000/2500 (Illumina): Reversible terminator, $2 \times 100$ bases, 11 days (regular mode) or 2 days (rapid run mode), $600 \, \text{Gb}$ (regular mode) or $120 \, \text{Gb}$ (rapid run mode), cost-effectiveness, massive throughput, short read lengths.
  • 5500xl SOLID (Life Technologies): Ligation, $75 + 35$ bases, 8 days, $150 \, \text{Gb}$, low error rate, very short read lengths.
  • PacBio RS (Pacific Biosciences): Real-time sequencing, 3,000 (maximum 15,000) bases, 20 minutes per day, simple sample preparation, very long read lengths, high error rate.
  • 454 GS Junior (Roche): Pyrosequencing, 500 bases, 8 hours, 0.035 Gb, long read lengths, high reagent costs, high error rate in homopolymers.
  • Ion Personal Genome Machine (Life Technologies): Proton detection, $100 - 200$ bases, 3 hours, short run times, high error rate in homopolymers.
  • Ion Proton (Life Technologies): Proton detection, up to 200 bases, 2 hours, short run times, high error rate in homopolymers.
  • MiSeq (Illumina): Reversible terminator, $2 \times 150$ bases, 27 hours, $1.5 \, \text{Gb}$, cost-effectiveness, short run times, read lengths too short for efficient assembly.

Genome Annotation

• Genome annotation involves identifying the elements within a genome.
• Not all of a genome encodes genes.
• For example:
  • E. coli is 70% protein-coding.
  • Humans only have 1.3% protein-coding regions.
• Other components include:
  • Simple repeats: Tandemly repeated units.
    • Microsatellites: $(1 - 7 \, \text{bp})$ e.g., (cacacacaacacaca….)
    • Minisatellites: Typically < $40 \, \text{bp}$
    • Satellites: $(140 - 360 \, \text{bp})$
  • Mobile elements: Transposable elements (>50% of human genome).
    • Parasitic stretches which spread through the genome.
    • Started as viruses, but most are now inactive.
    • 47 Types found in draft sequence of human genome.

Gene Structure - Prokaryotes

• Gene structure in prokaryotes includes regulatory regions, promoters, start codons, stop codons, and open reading frames.
• Genes to proteins flow.
• Start codon: AUG
• Stop codon: UGA

Gene Finding - Prokaryotes

• Prokaryotes are typically 60-70% coding.
• No splicing; therefore, one can look for large open reading frames (ORFs).
• May miss short genes.
• Which start codon to use? (ATG, TTG, GTG, ATT etc.).

Gene Finding - Eukaryotes

• Eukaryote nuclear genomes are much more complex than bacterial genomes.
  *

Gene Finding - Automation

• Due to the size of the task, manual annotation is impossible.
• Automated methods have been developed.
• Involves the differentiation of coding and non-coding regions.
• Identification of gene features (splice sites, promoters, regulatory elements, etc.).
• Content-based analysis, pattern recognition is used.

Gene Finding - Content Based Analysis

• Codon Preference: Species-specific preferences (e.g., GGA, GGT, GGC, GCG all encode Glycine, but some codons are used more than others).
• For each of the 6 possible frames of translation, go through the genome and calculate (using the codon bias table) the statistical likelihood that each codon is coding.

Gene Finding - Pattern Recognition

• Automated methods attempt to look for 'gene' features e.g. promoter regions, splice sites, ribosomal binding motifs etc..
• Example Prokaryotic promoter region
  • Nearly all Pribnow boxes have the three letters "TA***T"
  • *** is TAA in 50% of Pribnow boxes

Automated Genome Annotation

• Automated genome annotation is dependent on establishing homology with genes of known function; some database entries are wrong!
• Over $12,000$ users worldwide have annotated over $60,000$ distinct microbial genomes using RAST (census Jan 2014).

Gene Annotation

*Gene Annotation is an Ongoing Project

• The number of predicted ORFs and functionally assigned genes varies across genomes.

E. coli Genome

• Energy Allocation:
  • ~20% energy goes to small molecule metabolism.
  • ~12% energy goes to LARGE molecule metabolism
  • ~20% energy goes to cell structure & processes
• Operons: 2584 predicted & known operons; most (68%) have one promoter. Roughly 90% are thought to be regulated by only one protein.
  • 4,639,221 bp total.
  • 4288 protein-coding genes:
    • 30% "well characterized".
    • ~30% "no known function".
  • Average distance between genes: $118 \, \text{bp}$; (only 70 regions >600 bp).
  • Protein-coding genes account for ~88% of total.
  • ~1% stable RNAs.
  • ~1% repeats.
  • ~10% "regulatory".

E. Coli Genome Classification

• Gene classification based on COG functional categories (Translation, Transcription, DNA replication, etc.).

ECOCYC Database

• A member of the BioCyc database collection (Cellular Overview of Escherichia coli K-12 substr. MG1655).
• Genes of unknown function are not included in metabolic models.

Genome Projects

• The concept of the 'pan-genome':
  • Any one E. coli genome contains about 5000 genes, and roughly two-thirds of these are found in all E. coli genomes, but the other third are accessory genes, found in other strains, but not all.
  • The E. coli pan-genome consists of 90,000 genes.
  • Surprisingly, any one E. coli strain contains less than 10% of the total number of E. coli genes in the E. coli pan-genome.
  • The E. coli pan-genome is ~4x bigger than the human genome.
    *Accessory DNA is often found in 'genomic islands'.
• some of these have virulence genes and are termed 'pathogenicity islands' – derived from horizontal gene transfer

Repeat sequences

• Repeat sequences are abundant and variable.
• A huge diversity of transposable elements.
• Miniature Inverted-repeat Transposable Elements (MITEs):
  • less than $300 \, \text{bp}$, nonautonomous and do not transpose by themselves because they lack the transposase gene.
  • They appear to be the remnant of insertion sequences, with the terminal inverted-repeat (TIR) sequence, the direct repeats, and target site duplication.
• Clustered regularly interspaced short palindromic repeats (CRISPRs).

Metagenome Sequencing

• Only a small proportion of bacterial species can be cultured.
• NGS of environmental DNA helps build a picture of microbial diversity.
• For example, sequencing of DNA extracted from human stool samples. From the cohort of 124 European individuals, the first human gut microbial gene catalog was established as 3.3 million non-redundant genes
• The human gut microbial gene catalog can be correlated with specific disease conditions, e.g., to predict Type 2 Diabetes.

Functional Genomics

• Understanding gene function.
• Generating a metabolic model.
• Understanding regulatory networks.
• Understanding protein-protein interactions.
• Understanding complex biological machines.
• Understanding cell-cell interactions.
• Creating a molecular understanding of cell function.
• Involves "genome-wide," "global," "high-throughput," and "highly-parallel" approaches.
• Performed at multiple levels (experimental, computational).
• Data collection and analysis often require radically different technologies.
• Sometimes old methods can be scaled to a genomic level.

Functional Genomics: Analysis

• Transcriptomics (RNA-Seq): Compare expression of every gene for different growth conditions or between wild-type and regulatory mutant.
• Proteomics: Identify changes in expressed proteins/test protein-protein interactions.
• Global systematic mutagenesis: Identify function of annotated ORFs.
• Bioinformatics: Comparative genomics; analysis of vast data sets, etc.

Functional Genomics example: Streptomyces coelicolor

• A gram-positive filamentous soil bacterium.
• Genome fully sequenced at Sanger Institute using ordered cosmid library.
• Microarrays prepared and processed at UniS.
• Proteomics provided by John Innes Centre.
• Systematic genome mutagenesis using in vitro transposon mutagenesis of the ordered cosmid library at Swansea University.
• Select for marker replacement [AprRKanS] usually 1-10% of exconjugants if the gene/operon is non-essential.
• Transfer of cosmid carrying Tn insertion by conjugation from E.coli ET12567(pUZ8002) into S. coelicolor.
• Location and description of each insertion provided at http://strepdb.streptomyces.org.uk/.
• Bioinformatics integrates genome/microarray/proteomic/metabolomic/ gene function datasets - enabling better understanding of the biology of Streptomyces and subsequent exploitation to produce new pharmaceuticals in greater yields.

Genomics: Conclusions

• Next-gen sequencing has led to an exponential growth in genome data.
• Genome annotation can now be automated.
• Bacterial pan-genomes can be very large.
• Metagenomics is a tool to understand the human microbiome and can help predict disease.
• Functional genomics generates big data that can be integrated to generate a detailed description of the biology of bacteria.