L8 - Microarrays

Question 1

What are microarrays?

Answer 1

A substrate of some kind (eg. glass slide or quartz wafer) on which there are chemically bonded DNA probes

Works by hybridisation - Watson-Crick base pairing between probe & sample
• DNA/RNA in sample we want to bind is the TARGET

The target DNA/RNA is labelled with a fluorescent marker

Intensity of flouresence of feature relates to amount of target DNA in the sample – this allows us to observe changes for specific genes

Question 2

Microarray applications

Answer 2

Initial use was for gene expression, extract mRNA, convert to cDNA

Arrays run under different conditions, (drug vs control) allowed for differential gene expression as reported fold-changes

Allows us to see what genes are “switched-on” by the cell in response to different conditions, these can then be investigated or characterised

Question 3

Why do we care about gene expression?

Answer 3

The genes in a genome are not all actively transcribed at once

Detecting which mRNA molecules are around at any given moment in time shows which genes are being actively transcribed

Lowest level at which genotype gives rise to phenotype

Compare gene expression in normal Vs cancer cells / tissue

Can compare expression between different environmental conditions or stress to see how an organism responds

Question 4

Old methodology of microarrays

Answer 4

Northern blotting

Question 5

What is Northern blotting?

Answer 5

Northern Blot refers to capillary transfer of RNA from the electrophoresis gel to the blotting membranes

Principle: Northern blotting is a commonly used method to study gene expression by detection of RNA (or isolated mRNA) in samples

Question 6

Cons of northern blotting

Answer 6

Time consuming and laborious

Uses nasty chemicals and radioactive labelling
• Radioactive phosphate probes
• Formaldehyde

Can only work on a handful of genes

Provides mainly qualitative data

You need to decide in advance which genes you want to investigate

Question 7

Why are microarrays an improvement on northern blotting?

Answer 7

Less Laborious, just dump the sample over a chip, automated

Not radioactive

Can work on 10,000s of genes, can use genomes to design probe set

Has a much higher dynamic range enables relative abundances to be determined

Question 8

What is the difference between a northern blot and a western blot?

Answer 8

The northern blot is used chiefly to study gene expression by detection of RNA or RNA species (like mRNA)

The western blot is used to detect specific proteins in a sample by “immunoblotting”

Question 9

How do you measure microarrays?

Answer 9

Using an affymetrix

The little “windows” are the hybridisation chamber where the experimental samples are introduced

Question 10

Microarray method

Answer 10

1) Extract RNA from target
2) mRNA converted to cDNA via reverse transcriptase - labelled with fluorescent cyanine dye
3) Hybridisation to array (45-65 degrees)
4) Array has ssDNA probes bound in spots called features
5) Washing to remove non-specific binders
6) Laser scans of the array
7) Data processing

Question 11

How many base pairs make up a strand on the array?

Answer 11

25 bp

Question 12

How does hybridisation work?

Answer 12

Hybridisation between two DNA strands, the property of complementary nucleic acid sequences to specifically pair with each other by forming hydrogen bonds between complementary nucleotide base pairs

Shining with a laser caused flagged DNA fragments that hybridise to glow

Question 13

Why is it not easy to compare between 2 different microarrays directly using raw intensity?

Answer 13

Due to variation in multiple things:

1. Differences in sample extraction yield
2. Differences in hybridisation conditions, including background signal
3. Differences from one array to the next and artefacts on the array

Absolute abundance is difficult to quantify

Question 14

How do you reduce the variation seen using microarrays?

Answer 14

Replicates - preferably on the same day

Question 15

What are the 2 types of replicates?

Answer 15

Biological and technical

Biological are extra patients in a clinical study

Technical replicates would be using another array for the same patients sample, or even replicated samples

Question 16

How do you transform the intensity values?

Answer 16

Convert data to a log to the base of 2 (log2)

Will cover a smaller numerical range

Intensity will be bell shaped - normally distributed

Question 17

How do you convert transformed values back to the raw intensity values?

Answer 17

Anti-logging the values

Raise 2 to the power of X, where X is your log2 value of intensity

Question 18

What is the batch effect?

Answer 18

Batch effects show up where sample variability is due to sample handling, rather than genuine biological variation

Were samples extracted the same day?
Were samples hybridised on the same day?
Were sampled processed by the same person?

Batch effects can mask genuine results

Question 19

How can you spot batch effects?

Answer 19

Check how your samples cluster:
• Unsupervised hierarchical clustering successively links samples with similar expression profiles to give a tree like structure (dendrogram) and heatmap.
• And/or principal component analysis (PCA)

There are computational strategies for removing batch effects but it’s best not to introduce them in the first place

Question 20

What is PCA?

Answer 20

Principal Component Analysis
• PCA is a way of explaining variability in data.
• The first principal component describes the majority of the variability
• The second principal component accounts for as much of the remaining variability as it can and so on
• Plotting 3 principal components leads to a 3-dimension graph where alike samples should cluster together

A dimensionally reduction technique

Question 21

Analysis of PCA?

Answer 21

Allows you to spot and remove outlier arrays

We would expect that replicates all cluster together and and that different conditions separate out as clusters

Question 22

What is the main purpose of a gene expression microarray experiment?

Answer 22

To spot genes that are regulated by one or more of the experimental parameters:
• Time
• Drug / ligand response
• Stress, like oxidative or heat shock etc.
• Developmental process

Question 23

What happens if you have no replicates?

Answer 23

No statistics - can’t estimate population variance from sample size of one