Advanced Practical Genetics

Microarray - Hands-on data analysis

203.305

Dr. Pierre-Yves Dupont, Postdoctoral Researcher

p.y.dupont@massey.ac.nz

Planning

25.9.17 - 9-10am - AHB1.40B

Introduction (lecture)

2.10.17 - 9-10am - AHB1.40B

Paper discussion

3.10.17 - 10-1pm - C5.15

From raw data to lists of differentially expressed genes (Step by step analysis of a microarray data set using the R language)

9.10.17 - 9-10am C5.15

Lab discussion

10.10.17 - 10-1pm C5.15

Biological interpretation of microarray data (Gene ontology analysis using the R language + online research of candidate genes)

Microarray studies

  1. Introduction
  2. Microarray technology
  3. Analysis
  4. MIAME
  5. Examples of microarray studies (paper discussion topic and lab topic)

What are microarrays?

A microarray is a solid support (such as a membrane or glass microscope slide) on which DNA of known sequence is deposited in a grid-like array.

What are microarrays?

DNA microarray

What are microarrays

Hybridisation and transcriptomics?

The amount of RNA hybridised on each grid location can be measured and is a proxy for the gene expression level

Microarray applications

  • Gene expression analysis
  • Re-sequencing
  • SNP-analysis
  • DNA-Protein interactions

Expression Studies

Figure modified from: Katherine Joyce, Woods Hole Oceanographic Institution

Definitions

  • Genome: entire DNA sequence of an organism
  • Epigenome: chemical marks of the genome that modify its expression
  • Transcriptome: all gene transcripts present in a given cell/tissue at a given time (“snapshot”)
  • Transcriptomics: global analysis of gene expression = genome-wide expression profiling

Definitions

  • cDNA: complementary DNA made from mRNA by the enzyme reverse transcriptase
  • EST: Expressed Sequence Tag, small pieces of an expressed gene (cDNA)
  • Hybridisation: based on complementary molecules, sequences that are able to base-pair with one another. When two complementary sequences find each other, they will lock together, or hybridise (primer annealing, probe-target binding etc).

Genome-wide expression studies - Medical applications

  • Cancer research: Cell-cycle monitoring, genetic markers detection
  • Drug development and response: Treatment-induced expression pattern
  • Diagnosis: Disease-associated expression patterns

Genome-wide expression studies - Biological applications

  • Development biology: comparison of different developmental stages
  • Ecology: interactions between organisms (symbiosis, pathogenicity...) or between organisms and environment (temperature, nutrient...)
  • Evolution: within and between species variation, hybrids vs. parents, diploids vs. polyploids
  • Functional analyses: wild type vs. mutant

Microarray studies

  1. Introduction
  2. Microarray technology
  3. Analysis
  4. MIAME
  5. Examples of microarray studies (paper discussion topic and lab topic)

Microarray analysis principle

Image from: http://www.scq.ubc.ca/image-bank/

Microarray analysis principle

Image from: http://www.scq.ubc.ca/image-bank/

Competitive hybridisation

It is possible to represent different samples on one microarray using different fluorescent molecules (fluorophores)
  • Cyanin 3 (Cy3): green fluorescence (excited at 550nm, emission at 570nm)
  • Cyanin 5 (Cy5): red fluorescence (excited at 650nm, emission at 770nm)

Competitive hybridisation

Competitive hybridisation

Competitive hybridisation

Microarray study pipeline

Question driven

Goals? Hypothesis? Questions?

Microarray study pipeline

  • Platform/ design
    • What technology?
    • Source of the gene probes?
    • Cross-species hybridisation?
    • Replication level
    • Hybridisation scheme

Microarray study pipeline

  • Platform
  • Laboratory steps
    • Sample preparation and labelling
    • Hybridisation
    • Washing
    • Image acquisition

Microarray study pipeline

  • Platform
  • Laboratory steps
  • Bioinformatics steps
    • Data transformation and normalisation
    • Analysis of differentially expressed genes (statistical tests, gene ontology, ...)
    • Visualisation (graphics)
    • Data storage (databases, MIAME standards)

Microarray study pipeline

  • Platform
  • Laboratory steps
  • Bioinformatics steps
  • Data interpretation
    • Answers?
    • New hypotheses?
    • Follow-up experiments?
    • Validation?

Microarray studies

  1. Introduction
  2. Microarray technology
  3. Analysis
  4. MIAME
  5. Examples of microarray studies (paper discussion topic and lab topic)

Noise reduction

Before
After

Data normalisation

Global normalisation

Data normalisation

Lowess normalisation (LOcally WEighted Scatterplot Smoothing)

Expression ratios, M & A

  • $\color{green}{Cy3} = Sample1$ (Green)
  • $\color{red}{Cy5} = Sample2$ (Red)
  • $\color{red}{Cy5}$ > $\color{green}{Cy3}$: higher expression in sample 2
  • $\color{green}{Cy3}$ > $\color{red}{Cy5}$: higher expression in sample 1
  • Log fold ratio: $M = log_2({\color{red}{Cy5} \over \color{green}{Cy3}})$ = $log_2({\color{red}{Cy5}}) - log_2({\color{green}{Cy3}})$
  • Expression average: $A = {1\over2} (log_2(\color{red}{Cy5}) + log_2(\color{green}{Cy3}))$ = ${1\over2}log_2(\color{red}{Cy5} \color{green}{Cy3})$

Log Fold Ratio

Expression ratio: $log({\color{red}{Cy5} \over \color{green}{Cy3}})$

Log Fold Ratio

Reminder: $log_2(x)$ is the unique real number $y$ such that: $2^y = x$.
For example: $log_2(8) = 3$ because $2^3 = 8$

$\color{red}{Cy5}/\color{green}{Cy3}$ $log_2({\color{red}{Cy5}/\color{green}{Cy3}})$
4 2
2 1
1 0
0.5 -1
0.25 -2

Hypothesis testing

T-test

Null hypothesis ($H_0$): gene $x$ is not differentially expressed between two treatments
Mean:
$\overline{x}={{1 \over M} \sum_{i=1}^{M}{x_i}}$; for gene $x$ in M replicates
Variance:
$S_{x}^2 = {{1 \over {M-1}} \sum_{i=1}^{M}({x_i^2 - \overline{x}^2})}$
T-statistic:
$T_x={{ \overline{x_{C_1}} - \overline{x_{C_2}} } \over \sqrt[2]{ {S_{x_{C_1}}^2 \over M} + {S_{x_{C_2}}^2 \over N} } }$

T-test and P-value

T-test is used only to compare two samples. To compare more than two samples, ANOVA (ANalysis Of VAriance) is used.

T-test and P-value

Hypothesis testing

T-test

Null hypothesis ($H_0$): gene A is not differentially expressed between two treatments
  1. Compute the signal to noise ratio (difference of the means or medians) for each gene
  2. Compute the t-statistic for each gene using the replicates
  3. Compare t-statistic with the t-distribution
  4. If t-statistic is more extreme than the critical t-statistic at a chosen significance level (e.g. $\alpha=0.05$) reject the null hypothesis, otherwise accept it. P-value estimation

Quiz

Usually, a p <  0.05  is considered small enough to reject the null hypothesis of no biological effect in favour of the alternative hypothesis of a biological effect.
P-values are also known under type  1  error – the probability of rejecting the null hypothesis when it is actually true (= false positive rate).
P-value of 0.01 means a false positive rate of  1  %.
When analysing multidimensional data sets, p-values need to be adjusted for  multiple testing .
Two common p-value adjustment methods are  Bonferroni  and  False Discovery Rate .

Bonferroni Correction

  • If you hypothesize that a specific gene is up-regulated, p < 0.05 is fine.
  • If you hypothesize that any of 10,000 genes is up-regulated, with p < 0.05 you can expect to see 5% (500 genes) up-regulated by chance alone.
  • To account for the thousands of repeated measurements, some researchers apply a Bonferroni correction.
$p$ < $(0.05)/10,000$
$p$ < $5e^{-6}$
The Bonferroni correction is generally considered to be too conservative and False Discovery Rate (FDR) should be used.

False Discovery Rate

Benjamini-Hochberg method

Imagine an array with 6400 genes and an experiment where 184 genes are differentially expressed at $P=0.01$: 64 genes would be expected to appear differentially expressed by chance alone.

FDR = false discovery rate = ${{64}\over{184}} * 100 = 35 \%$

False Discovery Rate

Benjamini-Hochberg method

P-valueObserved number of genesExpected number of false positivesFDR
$10^{-2}$18464 35
$10^{-3}$ 35 6 18
$10^{-4}$ 150.6 4

With decreasing p-value, FDR also decreases, but so does the number of differentially expressed genes – choose a p-value which balances both!

Microarray studies

  1. Introduction
  2. Microarray technology
  3. Analysis
  4. MIAME
  5. Examples of microarray studies (paper discussion topic and lab topic)

MIAME Standard

Minimum Information About a Microarray Experiment that is needed to enable the interpretation of the results of the experiment unambiguously and potentially to reproduce the experiment
http://fged.org/Workgroups/MIAME/miame.html/

MIAME Standard

  1. Raw data for each hybridisation (CEL or GPR files)
  2. Processed (normalised) data (used to draw the conclusions from the study)
  3. Essential sample annotation including experimental factors and their values
  4. Experimental design including sample data relationships (e.g. which hybridisations are technical and biological replicates)
  5. Sufficient array annotation (e.g. gene identifiers, probe sequences)
  6. Essential laboratory and data processing protocols (e.g. normalisation method used to obtain the final data)

Gene expression databases

Gene Expression Omnibus (GEO) @NCBI (http://www.ncbi.nlm.nih.gov/geo/)

Gene expression databases

Geo Datasets @NCBI (http://www.ncbi.nlm.nih.gov/gds/)
Geo Profiles @NCBI (http://www.ncbi.nlm.nih.gov/geoprofiles/)

Gene expression databases

ArrayExpress @ EBI (http://www.ebi.ac.uk/arrayexpress/)

Gene expression databases

Expression Atlas @ EBI (http://www.ebi.ac.uk/gxa/)

Microarray studies

  1. Introduction
  2. Microarray technology
  3. Analysis
  4. MIAME
  5. Examples of microarray studies (paper discussion topic and lab topic)

Microarray paper discussion

Lab case study: Herbivory in Nicotiana attenuata (Solanaceae)

What type of research?
  • Which genes and metabolites defend plants against insects?
  • Costs and benefits of defense
  • Genetic engineering of defense traits
  • Plant pollination
MPI for Chemical Ecology, Baldwin Lab

Lab case study: Herbivory in Nicotiana attenuata (Solanaceae)

Why N. attenuata?
  • Diverse herbivore community
  • High plasticity (direct and indirect defense)
  • Easily cultivated annual species
MPI for Chemical Ecology, Baldwin Lab

Case study - Chips, veggies & vegetarians

Case study - Chips, veggies & vegetarians

The chip: cDNA array with 15,264 potato genes from TIGR (The Institute for Genomic Research)
The veggies The vegetarian
Solanum nigrum
Black nightshade		 Nicotiana attenuata
Coyote tobacco		 Manduca sexta
Question:
Do tobacco and black nightshade plants respond differently to caterpillar attack?

Microarray Case Study

RNA source

2 herbivore treatments and 2 controls

Microarray Case Study

6 arrays

Each arrow represents one array. Herbivore-induced tissue (Cy3) was co-hybridised control tissue (Cy5). Each comparison was replicated three times.

What will you do in the lab?

Lab 1
R warm-up exercise. Identification of differentially expressed genes
Lab 2
Identification of differentially expressed biological processes

Thanks

Slides available here:

http://bit.ly/massey_203305