1

Introduction to Bioinformatics and Gene Expression Technologies

Utah State University – Fall 2019Statistical Bioinformatics (Biomedical Big Data)Notes 1

2

Vocabulary

Gene:

Genetics:

Genome:

Genomics:

hereditary DNA sequence at a specific location on chromosome (that “does something”)

study of heredity & variation in organisms

an organism’s total genetic content (full DNA sequence)

study of organisms in terms of their genome

3

Vocabulary Protein:

Proteomics:

Phylogeny:

Phylogenetics:

Phenotype:

sequence of amino acids that “does something”

study of all of the proteins that can come from an organisms’ genome

the evolutionary or historical development of an organism (or its DNA sequence)

the study of an organism’s phylogeny

the physical characteristic of interest in each individual – for example, plant height, disease status, or embryo type

4

Vocabulary Bioinformatics:

Statistical Bioinformatics:

the collection, organization, & analysis of large-scale, complex biological data

the application of statistical approaches to bioinformatics, especially in identifying significant changes (in sequences, expression patterns, etc.) that are biologically relevant (especially in affecting the phenotype, or in response to some treatment)

5

RNA-Seq Example: 8 patients, 56,621 genes 8 heart tissue samples 4 control (no heart disease) 4 cardiomyopathy (heart disease) 2 restrictive (contracts okay, relaxes abnormally) 2 dilated (enlarged left ventricle)

These “Naples” data made public by Institute of Genetics and Biophysics (Naples, Italy)

Ctrl_3 RCM_3 Ctrl_4 DCM_4 Ctrl_5 RCM_5 Ctrl_6 DCM_6ENSG00000000003 308 498 362 554 351 353 220 309ENSG00000000005 3 164 2 43 13 83 22 16ENSG00000000419 1187 1249 1096 1303 970 863 637 684ENSG00000000457 163 239 168 195 153 194 44 117ENSG00000000460 63 108 83 109 87 43 54 51ENSG00000000938 369 328 272 669 1216 193 861 292...

url <- "http://www.stat.usu.edu/jrstevens/bioinf/naples.csv"naples <- read.csv(url, row.names=1)

Common statistical research objectives Test each gene (row) for “differential

expression” between conditions Ctrl vs. non-Ctrl Dilated vs. Restrictive Restrictive vs. Ctrl etc.

Test specific groups of genes (with a known common function) for overall expression differences between conditions Which functions are differentially active between

Ctrl and non-Ctrl, for example?6

HW data: bovine oviduct“Effect of lactation and location relative to the corpus luteum on the transcriptome of the bovine oviduct epithelium”

Biology prerequisite: opposable thumb(s) Why does context matter? Interdisciplinary collaborations require

communication, curiosity, and learning new fields Statistical collaborator vs data analyst

(2019 Vance & Smith ASCCR framework) NCBI GEO database: GSE124110

7url <- "http://www.math.usu.edu/jrstevens/bioinf/GSE124110_readCounts_oviduct.txt"bovine <- read.table(url, header=TRUE)

8

Central Dogma of Molecular Biology

9

A road map to bioinformaticsCentral Dogma Technology

GenomicHypothesis

Type of Study or Analysis

Gene GenomeSequencing

Genotype QTL

mRNAtranscript

TranscriptProfiling Transcriptome

Microarrays or Next-Gen Sequencing(Epigenetics / methylation)

Protein Protein quantification and function

Proteome Protein Microarrays or Proteomics

Phenotype(From introductory lecture by RW Doerge at 2013 Joint Statistical Meetings)

10

“Alphabets”

DNA sequences defined by nucleotides (4)

DNA sequence mRNA sequenceProtein sequence

Protein sequences defined by amino acids (20)

11

General assumption of gene expression technology

Use mRNA transcript abundance level as a measure of the level of “expression” for the corresponding gene in the biological sample

Proportional to degree of gene expression

Side note: a “methylated” gene is silenced (no expression)

12

How to measure mRNA abundance? Several different approaches with similar

themes: Affymetrix GeneChip Nimblegen array Two-color cDNA array More modern: next-generation sequencing (NGS)

Representation of genes on slide Small portion of gene (“oligo”) Larger sequence of gene Blank slate (NGS)

oligonucleotide arrays

13

General DNA sequencing Sanger

1970’s – today most reliable, but expensive

Next-generation [high-throughput] (NGS): Genome Sequencer FLC (GS FLX, by 454

Sequencing) Illumina’s Solexa Genome Analyzer Applied Biosystems SOLiD platform others … Key aspect: sequence (and identify) all sequences

present

14

Common features of NGS technologies (1)

fragment prepared genomic material biological system’s RNA molecules RNA-Seq

DNA or RNA interaction regions ChIP-Seq, HITS-CLIP

others …

sequence these fragments (at least partially) produces HUGE data files (~10 million

fragments sequenced)

15

Common features of NGS technologies (2)

align sequenced fragments with reference sequence usually, a known target genome (gigo…) alignment tools: ELAND, MAQ, SOAP, Bowtie,

others often done with command-line tools still a major computational challenge

count number of fragments mapping to certain regions usually, genes these read counts linearly approximate target

transcript abundance

16

Here, RNA-Seq: recall central dogma:

DNA mRNA protein action quantify [mRNA] transcript abundance

Isolate RNA from cells, fragment at random positions, and copy into cDNA

Attach adapters to ends of cDNA fragments, and bind to flow cell (Illumina has glass slide with 8 such lanes –so can process 8 samples on one slide)

Amplify cDNA fragments in certain size range (e.g., 200-300 bases) – using PCR clusters of same fragment

Sequence – base-by-base for all clusters in parallel https://www.youtube.com/watch?v=-7GK1HXwCtE

17(originally illumina.com download)

18(originally illumina.com download)

19(orginally illumina.com download)

20(orginally illumina.com download)

21

Then align and map … For sequence at each cluster, compare to [align

with] reference genome; file format: millions of clusters per lane approx. 1 GB file size per lane

For regions of interest in reference genome (genes, here), count number of clusters mapping there requires well-studied and well-documented

genome

22

A short word on bioinformatic technologies “Never marry a technology, because it will

always leave you.” – Scott Tingey,

Director of Genetic Discovery at DuPont (shared in RW Doerge 2013 introductory overview lecture at 2013 JSM)

In this class, we will discuss only a couple of technologies, emphasizing their recurringstatistical issues These are perpetual (and compounding)

A Rough Timeline of Technologies (1995+) Microarrays

require probes fixed in advance – only set up to detect those (2005+) Next-Generation Sequencing (NGS)

typically involves amplification of genomic material (PCR)(can bias GC-rich regions; also may lead to misassemblies and gaps [2018])

(2010+) Third-Generation Sequencing “next-next-generation” – Pac Bio, Ion Torrent no amplification needed – can sequence single molecule longer reads possible; still (2013 ; 2016) showing high errors

(2012+) Nanopore-Based Sequencing [very promising] Oxford Nanopore, Genia, others bases identified as whole molecule slips through nanoscale

hole (like threading a needle); coupled with disposable cartridges; still (2013 ; 2016) under development (2018)

(?+) more …23

Differ in how

sequencing done; subsequent post-alignm

ent statistical analysis basically same

24

Affymetrix Technology – GeneChip

Each gene is represented by a unique set of probe pairs (usually 12-20 probe pairs per probe set)

Each spot on array represents a single probe (with millions of copies)

These probes are fixed to the array

(Image courtesy Affymetrix, www.affymetrix.com)

25

Affymetrix Technology – Expression

(Images courtesy Affymetrix, www.affymetrix.com)

A tissue sample is prepared so that its mRNA has fluorescent tags; wait for hybridization; scan to light tag

26

Affymetrix GeneChip

Image courtesy Affymetrix, www.affymetrix.com

27

Cartoon Representations (originally from Affymetrix outreach)

Animation 1: GeneChip structure(1 min.)

Animation 2: Measuring gene expression(2.5 min)

28

Images; Affymetrix data is probe intensity

Images courtesy Affymetrix, www.affymetrix.com

Full Array Image Close-up of Array Image

29

How to analyze data meaningfully? Consider (for any technology):

Data quality Data distribution Data format & organization Appropriateness of measurement methods (& variance) Sources of variability (and their types) Appropriate models to account for sources of variability and

address question of interest Meaning of P-values and appropriate tests of significance Statistical significance vs. biological relevance Appropriate and useful representation of results

Many useful tools available from Bioconductor

30

The Bioconductor Project

“Bioconductor is an open source and open development software project for the analysis and comprehension of genomic data”

Not just for RNA-Seq or microarray data

Like a living family of software packages, changing with needs

Core team mainly at Fred Hutchinson Cancer Research, plus many other U.S. and international institutions

Source: www.bioconductor.org

31

Main Features of the Bioconductor Project

Use of R Documentation and reproducible research Statistical and graphical methods Annotation Short courses Open source Open development

Source: www.bioconductor.org

32

What will we do in this class?

Learn basics of a few major Bioconductor tools

Focus on statistical issues

Discuss recent developments

Learn to discuss all of this in context