Term
| What are the reasons to perform transcriptome analysis? |
|
Definition
• mRNA (miRNA etc.) as an efficient
readout of a cell state (i.e. Phenotype)
• Goal is to quantify all mRNA molecules in
a cell or population of cells |
|
|
Term
| What are two DNA array technologies? |
|
Definition
|
|
Term
| What is the relationship between the probes and sample? |
|
Definition
Probe needs to be in excess!
Microarray: Probe is fixed on surface, sample is labeled |
|
|
Term
| What are the features of qualitative RNA-seq? |
|
Definition
• Goal is a comprehensive description of RNA
molecules
• Reads are assembled into mRNA molecules
• With or without genome
• Issues: 1) Even coverage 2) Long reads 3) Strand-specificity 4)Lot of data per sample |
|
|
Term
| What are the features of quantitative RNA-seq? |
|
Definition
• Goal is linking a variable (e.g. genotype, treatment) to gene expression patterns
• Reads are counted per gene/transcript
• Genome annotation necessary
• Issues: Little technical variation,Read length for mapping,Low library costs,Lot of samples |
|
|
Term
| Main features of microarrays |
|
Definition
• Probes need to be physically available
• Established technology
• Can interrogate a defined subset
• Fixed costs per sample
• Hybridisation background makes it impossible to estimate false negatives
• No distinction between transcripts of similar sequence |
|
|
Term
|
Definition
• Genome or transcriptome annotation necessary
• Developing technology
• Special technology is needed to target subsets
• Flexible cost per sample via barcoding, now comparable to array
• No background if reads can be unambigously mapped
• Even alleles can be distinguished |
|
|
Term
| What are the functions of normalization? |
|
Definition
•remove trivial factors (more label, more RNA, more sequencing depth etc., backgrounds on arrays)
• Usual assumption is that the total mRNA amounts are the same in all samples |
|
|
Term
| What are the functions of transformation? |
|
Definition
• Goal: make distributions of expression values suitable for normality assumptions
• Several different transformations exist
• Most common/easiest is the log transformation
• Fold-changes matter rather than absolute changes
• Log2 is convenient since a 2-fold-change is a difference of one |
|
|
Term
| What are the different types of transcriptome analyses? |
|
Definition
• Differential analysis/marker selection
• Class discovery (unsupervised learning)
• Class prediction (supervised learning)
• Pathway analysis |
|
|
Term
| What is the condition for differential analysis/marker selection? |
|
Definition
| For each gene ask whether its mean expression level is the same (null hypothesis) or different) |
|
|
Term
| What tests should we use for differential analysis/marker selection? |
|
Definition
• Parametric: T-test / ANOVA
• Non-parametric tests: Wilcoxon rank test,Mann-Whitney U test |
|
|
Term
| PINGO: Which statements are true regarding the processing of expression data? |
|
Definition
+If expression data is Log2-transformed and for Gene X sample A has a value of 5 and sample B has a value of 7, Gene X is expressed four-fold more in sample B
+If expression data is Log2-transformed and for Gene X sample A has a value of 5 and sample B has a value of 7, Gene X is expressed four-fold more in sample B
( change is symmetrical) |
|
|
Term
| PINGO: Which statements are true regarding differences between microarrays and RNA-Seq for the analysis of transcriptomes |
|
Definition
+Allele-specific expression is difficult to analyse using microarrays ( they rely on specificity of hybridization)
+The specificity of the detection relies on hybridization in the case of microarrays and on mapping reads to the genome in the case of RNA-Seq( in RNA this problems controlled better, know false negative rate)
+To design microarrays one needs the sequence information for a transcriptome
+Unknown transcripts are difficult to discover using microarrays |
|
|
