REGULOME – Lecture 2
COO1
p-value en log fold change bepalen drempel vanaf wanneer het differential gene expression is
Always statistical significant
p-value (adjusted for multiple testing) < 0.05
P-value (adjusted, FDR) < 0.1
Fold Change (log2 scale): how much change do I consider biologically relevant?
>1 or <-1
Too little genes --> accept to 0.6 (50% op or 25% down)
Ignore it completely --> 0
Amount of genes
Between 200 and 2000
Good for analysis with Gene Ontology or pathway analysis
Pathway analysis seperately for up- and downregulated genes!
RNA sequencing, transcriptome and
expression quantification
Part 2: RNA-seq (RNA-seq data analysis)
Analyse whole transcriptomes rather efficiently
Start with RNA, not every RNA species that is in the cell (95% is ribosomal RNA)
How to get rid of ribosomal RNA:
o Fish out with probes
o Select only poly-A mRNAs
Fragmentation to short pieces (enzymes, heat) to 100s of bp
Reverse Transcription --> cDNA = sequence library
(Eerst RT dan frag of omgekeerd)
Sequence library has short, random fragments of RNA
, 1: preselect mRNA species,
2: short sequence reads from random parts of mRNAs
After analysis try to reconstruct original transcripts
Benefits (vs. Microarray)
Independence on prior knowledge (just sequence) --> discover new things
High resolution (every single nucleotide is sequenced), good sensitivity (efficient library
preparation, aslo lowly expressed genes), large dynamic range (no problems with highly
expressed genes, ratios reflect real abundance well)
Unravel previously inaccesible complexities (in transcripts)
Challenge
Interpretation is not straightforward
Procedures continue to evolve, no good standards (how many reads?)
Applications (what can you detect?)
Differential gene expression
Gene fusions
Alternative splicing
Novel transcribed genes (no prior knowledge)
Allele-specific expression (one variant more abundantly expressed)
RNA editing (RNA not 100% stable, some bases can be edited later)
Transcriptome for non-model organisms (sequence any organisms you want without
knowing its genome)
From reads to differential expression
Raw sequence data (FASTQ files) --> quality check
Reads mapping. QC for mapped reads
COO1
p-value en log fold change bepalen drempel vanaf wanneer het differential gene expression is
Always statistical significant
p-value (adjusted for multiple testing) < 0.05
P-value (adjusted, FDR) < 0.1
Fold Change (log2 scale): how much change do I consider biologically relevant?
>1 or <-1
Too little genes --> accept to 0.6 (50% op or 25% down)
Ignore it completely --> 0
Amount of genes
Between 200 and 2000
Good for analysis with Gene Ontology or pathway analysis
Pathway analysis seperately for up- and downregulated genes!
RNA sequencing, transcriptome and
expression quantification
Part 2: RNA-seq (RNA-seq data analysis)
Analyse whole transcriptomes rather efficiently
Start with RNA, not every RNA species that is in the cell (95% is ribosomal RNA)
How to get rid of ribosomal RNA:
o Fish out with probes
o Select only poly-A mRNAs
Fragmentation to short pieces (enzymes, heat) to 100s of bp
Reverse Transcription --> cDNA = sequence library
(Eerst RT dan frag of omgekeerd)
Sequence library has short, random fragments of RNA
, 1: preselect mRNA species,
2: short sequence reads from random parts of mRNAs
After analysis try to reconstruct original transcripts
Benefits (vs. Microarray)
Independence on prior knowledge (just sequence) --> discover new things
High resolution (every single nucleotide is sequenced), good sensitivity (efficient library
preparation, aslo lowly expressed genes), large dynamic range (no problems with highly
expressed genes, ratios reflect real abundance well)
Unravel previously inaccesible complexities (in transcripts)
Challenge
Interpretation is not straightforward
Procedures continue to evolve, no good standards (how many reads?)
Applications (what can you detect?)
Differential gene expression
Gene fusions
Alternative splicing
Novel transcribed genes (no prior knowledge)
Allele-specific expression (one variant more abundantly expressed)
RNA editing (RNA not 100% stable, some bases can be edited later)
Transcriptome for non-model organisms (sequence any organisms you want without
knowing its genome)
From reads to differential expression
Raw sequence data (FASTQ files) --> quality check
Reads mapping. QC for mapped reads
