Gene Ontology and overrepresentation analysis

Transcription

1 Gene Ontology and overrepresentation analysis Kjell Petersen J Express Microarray analysis course Oslo December 2009 Presentation adapted from Endre Anderssen and Vidar Beisvåg NMC Trondheim

2 Overview How can ontologies (and pathway) information help us What is an ontology? The Gene Ontology and how it's structured How to use Interactively Statistically Additional things to think about

3 So here you are Figure of diff exp

4 Gene lists Long list of differentially expressed genes Possibly hundreds of papers describing the functions of the genes Misleading names Different names in dfifferent organisms

5 What s in a name? The same name can be used to describe different concepts What is a cell?

6 Cell

7 Cell

8 Cell Image from

9 Gene Ontology (GO) Ontologies Sequence Ontology (SO) (sequence features) Phenotype and Trait Ontology (PATO) Taxon (NCBI) Anatomy (Penn) Disease (ICD9) Developmental stage (multiple sources)

10 Why Gene Ontology? Gene Ontology (GO) Produce a controlled vocabulary describing aspects of molecular biology, that can be applied to all organisms. Facilitate communication between people and organization. Improve interoperability between systems.

11 Goal of GO Consortium ( Produce a controlled vocabulary describing aspects of molecular biology, that could be applied to all organism. Describe gene products using vocabulary terms (annotation). Develop tools: to query and modify the vocavularies and annotations

12 How does GO work? What information might we want to capture about a gene product? What does the gene product do? Why does it perform these activities? Where does it act?

13 The Gene Ontology (GO) Molecular function: Gene product at biochemical level. Biological process: Cellular events to which the gene product contributes. Cellular component: Location or complex of gene/protein.

14 Molecular Function activities or jobs of a gene product Insulin binding insulin transport activity

15 Molecular Function drug transporter activity

16 Biological Process a commonly recognized series of events cell division

17 Cellular Component where a gene product acts

18 Content of GO Molecular Function 7,309 terms Biological Process 10,041 terms Cellular Component 1,629 terms Total 18, 975 terms Obsolete terms: 992 As of October 2005

19 Ontology Structure Directed acyclic graphs (DAGs) Relationships is a a is a type of b (e.g. truck is a car, or mitochondrion is an organelle) Regulates Positively regulates Negatively regulates part of sub process of (process) physical part of (component) (e.g. engine is part of a car, or mitochondrion membrane is a part of a mitochondrion)

20

21 Term Definitions and Curation The definitions for each GO term are being primarily derived from the Oxford Dictionary of Molecular Biology, or from relevant literature sources (SWISS PROT, PIR, NCBI CGAP, EC...). Curators around the world shifting through genomic and proteomic data then use the definitions and GO terms provided by GO to annotate or curate the genes and proteins in their favorite species. GO is stored as flat files, as XML files and as a relational database implemented in MySQL.

22 GO Annotation Association between gene product and applicable GO terms Provided by member databases. Collaborating databases annotate their gene products (or genes) with GO terms, providing references and indicating what kind of evidence is available to support the annotations. Made by manual or automated methods. GO Annotation Database object: gene or gene product GO term ID Evidence supporting annotation Reference publication or computational method

23 Gene Ontology and Microarrays Hypothesis: Functionally related, differentially expressed genes should accumulate in the corresponding GO group. Problem: Find a method, which scores accumulation of differential gene expression in a node of the Gene Ontology. GO tools can be important in order to answer questions such as: are genes involved in process P overrepresented among the total of differentially expressed genes in an experiment or does treatment A induce more genes involved in process P than treatment B?".

24 Browsing GO in J Express

25 Overrepresentation of GO terms We have a subset of genes List of differentially expressed genes List of genes that cluster together Which biological processes do these genes take part in? Is there an over representation of the number of genes belonging to a particular biological process, compared to what could be expected?

26 Question If we look at the dataset containing all of our genes and see that 10% of these belong to cell cycle. We then do a differentially expressed genes analysis and get a list of genes we believe are significantly changed. How many of the genes in the gene list do you expect belong to cell cycle?

27 Setup We name our subset of interesting genes for test data Reference data And the dataset containing all of our genes, the dataset we extracted the interesting genes Test datafrom and that we want to compare our testdata to, for reference data

28 Gene Ontology Analysis Reference data Test data Statistical comparison between the two GO components

29 Some gotchas P value not corrected for multiple testing Bonferroni correction: multiply by number of terms mapped to your dataset (sum of 3 blue numbers at the 3 top terms) Results depend on Selected cut off for test data (check for consistency over several cut offs) Version of GO obo and association files (keep track of which used) Browse the terms in the neighbourhood of high scoring terms, will often reveal more context.

30 Nice to know Alternative term hierarchies / tools GO slim Panther DB ( DAVID (