Today s Lecture Topics q Whole genome sequencing q

Today’s Lecture Topics q Whole genome sequencing q Shotgun sequencing method q Sequencing the human genome q Functional/comparative genomics q Transcriptome & RNA-Seq q Proteomics

Shotgun DNA sequencing: q Sequence the entire genome rapidly. q No requirement for a high resolution linkage or physical map. q Just break the genome up into small pieces, sequence it, assemble, and find the gene of interest/do the bioinformatic analysis later. q Reverses the way genetic studies proceeds. q It used to be we had to find the gene first to study the cause of the disease. q Now we can study genes we didn’t even knew exist.

Fig. 8. 13, Shotgun sequencing a genome

Shotgun DNA sequencing---dideoxy method: 1. Begin with genomic DNA and/or 200 -300 kb BAC clone library. 2. Mechanically shear DNA into ~2 kb bp overlapping fragments. 3. Isolate on agarose, purify, and clone into standard plasmid vectors. 4. Sequence ~500 bp from each end of each 2 kb insert. 5. Sequence from the middle 1, 000 bp of each insert is obtained from overlapping clones. 6. Repeat the process so that 4 -5 x the total length of the genome is sequenced (dideoxy sequencing is 99. 99% accurate). 7. Results in a contig library with ~97% genome coverage (the missing 3% is composed mostly of repeated DNA sequence). 8. Assemble hundreds of thousands of overlapping ~500 bp sequences with fast computers operating in parallel (supercomputer).

How to deal with the repeated DNA - 2 kb clones present a problem, solved with 10 kb clones: 1. Many repeated sequences in the genome are in regions spanning ~5 kb in size. 2. So many 2 kb clones contain entirely repeated DNA. 3. Results in a dead stop in the assembly, because there is ambiguity about where each clone goes. ü Repeated sequences occur all over the genome. 4. On average, 10 kb clones contain less repeated DNA sequence. 5. Solution is to create and sequence a 10 kb clone library derived from the same genomic DNA or BAC library. 6. Complete genome coverage requires combining the sequences from the 2 kb & 10 kb libraries.

Genome Date Size Institute Homo sapiens mt. DNA 1981 Haemophilus influenzae (bacteria) 1995 1, 830, 137 bp TIGR (1 circular) Shotgun Mycoplasma genitalium (bacteria) 1995 580, 070 bp TIGR (1 circular) Shotgun 16, 159 bp (1 circular) Method - Escherichia coli 1997 (bacteria) 4, 639, 221 bp University of (1 circular) Wisconsin. Madison Shotgun Methanococcus jannaschii (Archaeon) 1, 739, 933 bp DOE (3 circular) Shotgun 1996 Saccharomyces 1996 cerevisiae (yeast) 12, 067, 280 bp 100+ labs (16 linear) Mapping Caenorhabditis elegans (nematode) 97, 000 bp Consortium (6 linear) Mapping 1998

Genome Date Size Institute Drosophila melanogaster (fruit fly) 2000 180, 000 bp UC Berkley Celera Genomics Arabidopsis thaliana (angiosperm) 2000 125, 000 bp Consortium (5 linear) Homo sapiens (human) 2000 3, 400, 000 bp Human Genome Project & Celera Genomics Method Shotgun w/BAC map Mapping & Shotgun

Sequencing the human genome: Two major players: Human Genome Project (HGP): ü ü Publicly funded international consortium (NIH, DOE, etc. ) Francis Collins, National Human Genome Res. Inst. (NHGRI) Began in U. S. in 1990 with a goal of 15 years Genetic and physical mapping approach + dideoxy sequencing Celera Genomics Corporation (CRA): ü ü ü Spin-off of Applied Biosystems (ABI) J. Craig Venter, CEO Created in 1998 with a goal of 3 years Direct shotgun approach + dideoxy sequencing (+ HGP’s maps for validation) Both groups collected blood and sperm samples from anonymous male and female donors of different ethnic backgrounds.

J. Craig Venter Celera Genomics Francis Collins Human Genome Project

Milestone: 26 June 2000 White House press conference with Bill Clinton: HGP: Started 1990 ~22. 1 billion nucleotides of sequence data 7 -fold coverage Unfinished (24% completely finished, 50% near-finished) Celera: Started 1998 ~14. 5 billion nucleotides of sequence data 4. 6 -fold coverage Complete assembled genome with >99% coverage First assembled draft of human genome simultaneously published in Nature & Science 15 & 16 February 2001 (Nature published 1 day earlier).

How did Celera et al. assemble the sequences using shotgun methods? Method A: 1. Assembly of 26. 4 million 550 bp sequences 4. 6 -fold coverage, without reference to a physical map of any kind. 2. Covered >99% of the genome. 3. 500 million trillion base-to-base comparisons. 4. 20, 000 CPU hours (833 CPU days) on a year 2000 supercomputer. Method B: 1. Used BAC clone scaffold (combined lots of smaller maps) to validate the whole genome direct shotgun assembly approach. 2. Also helped resolved ambiguities resulting from the assembly of short repeated DNA fragments.

Features of the human genome: ü 32, 000 genes estimated (50, 000 -100, 000 were predicted). ü Not many more genes than Drosophila, and only 50% more genes than Caenorhabditis elegans (nematode worm). ü Only 1 -1. 5% of the genome codes for protein. ü 50% of the sequence is repeated DNA. ü Humans share 223 genes found in bacteria, but not yeast, nematodes, or fruit flies.

First human genome required $1 Billion USD and 13 years (or 2 years by Celera shotgun sequencing). Now it requires only a couple $1000 s and is done in 2 days.

Next-generation shotgun genome sequencing: q The shotgun method is fundamentally the same, but uses pyrosequencing and shorter read lengths (~150 bp paired-ends on Illumina). q 300 -800 bp fragments + mate-pairs of 2 -12 kb to aid assembly and increase N 50 (avg. scaffold length). q The throughput has increased and the cost has decreased. q Not uncommon to assemble trillions of sequence reads. Some things to consider: If error rates are high (454, Illumina) 30 -50 x genome sequencing is required to get a good genome. If error rates are low (SOLi. D, Ion Torrent) 4 -5 x coverage is sufficient. q Costs have been falling from $10 K to $1 K.

Sequence Contig Scaffold (contig of contigs)

Scaffold with two small gaps and one large gap bridged by mate pair with paired-end sequencing xxxxx---------------xxxxx xxxxx--------xxxxx xxxxx---------------------------xxxxx xxxxx---------------xxxxx xxxxx--------xxxxx mate pair Larger the scaffolds, larger the N 50 s

Sequencing is no longer the primary need; data storage/retrieval and computational needs are outpacing everything else. How much data storage does 1 human genome require? About 1. 5 GB (2 CDs) if your stored only one copy of each letter. For the raw format containing image files and base quality data 2 -30 TB are required. 30 -50 x coverage requires more data storage capacity. Sequence + quality scores is compressed to format called FASTQ. @SEQ_ID GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT + !''*((((***+))%%%++)(%%%%). 1***-+*''))**55 CCF>>>>>>CCCCCCC 65

FASTQ '!' represents the lowest quality while '~' is the highest. Left-to-right increasing order of quality (ASCII 90 characters): !"#$%&'()*+, -. /0123456789: ; <=>? @ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_`abcdefghijklmnopqrstuvwxyz{|}~ Illumina Sequence Identifiers @EAS 139: 136: FC 706 VJ: 2: 2104: 15343: 197393 1: Y: 18: ATCACG Bar code at front of each sequence that allows you to label each sequence (e. g. individual, population, etc. )

Sequence assembly & genotyping • Trimming and filtering sequences based on base quality scores Aligning reads to a reference genome Genotyping to determine homozygous & heterozygous SNPs http: //gatkforums. broadinstitute. org/

Post-genome sequencing era is very different: ü Classical genetics studies started with a phenotype and set out to identify the gene. ü But we now have the ability to start with a complete genome and set out to identify the phenotype. ü Large data sets required many computational and mathematical tools, which requires strong bioinformatics skillsets (Perl, Python, R, etc. ). Lots of applications: 1. Identify genes within genomic DNA sequences. 2. Align and match homologous gene sequences in databases and seek to determine function. 3. Predict structure of gene products. 4. Describe interactions between genes and gene products. 5. Study gene expression.

1. Identifying genes in DNA sequences: ü First step is annotation = identification and description of putative genes and other important sequences. ü Open reading frames (ORFs) ORF = potential protein coding sequence that begins with a start codon and ends with a stop codon. ü ORFs come in all sizes. ü Not all ORFs encode proteins (6 -7% do not in yeast). ü ORFs with introns can require sophisticated computer algorithms to detect (especially if there are many introns or introns are particularly long).

2. Homology searches to assign gene function: ü Homology search = identify gene function by searching database. ü Similarities reflect evolutionary relationships and shared function. ü Homology searches are performed for nucleotides and amino acids using BLAST = Basic Local Alignment Search Tool. ü Gen. Bank’s BLAST site: http: //www. ncbi. nlm. nih. gov/BLAST/ ü Example, human mt. DNA control region sequence: § TTCTCTGTTCTTCATGGGGAAGCAGATTTGGGTACCACCCAAGTATTGACTCACC CACAACAACCGCTATGTATTTCGTACATTACTGCCACCATGAATATTGCAC GGTACCATAAATACTTGACCACCTGTAGTACATAAAAACCCAATCCACATCAAAA

(2006)Jiawei Han & Micheline Kamber

Fig. 9. 2, Summary of genes in the yeast genome.

3. Gene function can be identified and studied in other ways: ü Gene knockout approach = systematically delete different genes and observe the phenotypes (PCR + cloning is one method). ü Synthesize recombinant proteins with modified amino acid sequence and expressed in E. coli. ü Test effects of mutations that don’t exist in nature.

ü Study the transcriptome = complete set of m. RNAs in a cell ü m. RNAs are not stable, but types and levels change with different experimental conditions. ü Sample m. RNA at experimental intervals and convert to c. DNA using reverse transcriptase. 1. Probe unknown c. DNAs with DNA microarray of PCR-generated ORF sequences (requires known sequence for each probe). 2. Or better yet, sequence the entire transcriptome using: RNA-Seq = Whole Transcriptome Shotgun Sequencing of all expressed RNAs. 3. Sequencing of ribosome-bound m. RNA for monitoring in-vivo translation.

Fig. 9. 7 b, Microarray study of gene expression

http: //www. nature. com/nbt/journal/v 28/n 5/images_article/nbt 0510 -421 -F 1. gif

The ribosome profiling strategy for monitoring translation in vivo by deep sequencing of ribosome-protected m. RNA fragments Nicholas T Ingolia, Gloria A Brar, Silvia Rouskin, Anna M Mc. Geachy & Jonathan S Weissman Nature Protocols 7, 1534– 1550 (2012) doi: 10. 1038/nprot. 2012. 086

“Proteomics”: Proteome = complete set of expressed proteins in a cell Major goals of proteomics: • Identify every protein, isolate and purify. • Determine the sequence and structure of each protein (and its function). • Create a database with the sequence of each protein. • Analyze protein levels and interactions in different cell types, at different times, and at different stages of development. Rationale: ü Genes are two-steps removed from disease (DNA m. RNA protein). ü Most gene products involved in disease are composed of protein. ü Understanding protein means understanding disease.

http: //biol. lf 1. cuni. cz/ucebnice/en/proteomics. htm

“Systems Biology” Computational and mathematic modeling of complex biological systems---Wikipedia Requires integration of genomic, proteomic, and metabolic data.