1	Spring 2011 BMD6621 – High-Throughput Sequencing Analysis Data Integration Shu-Jen Chen, Ph.D. Department of Biomedical Sciences Chang Gung University Jun. 3, 2011 (Friday 8:30 – 12:00)
2	To fully utilize the results of contemporary biological research, one would like to analyze data on biological function in addition to sequence information.
3	Unfortunately … Compared to sequence information, biological function is much more difficult to analyze. Biological data is fragmented Biologists currently waste a lot of time and effort in searching for all of the available information about each small area of research. Language used in biological research is not well controlled This is hampered further by the wide variations in terminology that may be common usage at any given time, which inhibit effective searching by both computers and people.
4	A simple example If you were searching for new targets for antibiotics, you might want to find all the gene products that are involved in bacterial protein synthesis, and that have significantly different sequences or structures from those in humans. If one database describes these molecules as being involved in 'translation‘ while another uses the phrase 'protein synthesis', it will be difficult for you - and even harder for a computer - to find functionally equivalent terms.
5
6
7	The Gene Ontology
8
9
10
11	What is the Gene Ontology project? The Gene Ontology (GO) project is a collaborative effort to address the need for consistent descriptions of gene products in different databases. The project began as a collaboration between three model organism databases, FlyBase (Drosophila), the Saccharomyces Genome Database (SGD) and the Mouse Genome Database (MGD), in 1998. Since then, the GO Consortium has grown to include many databases, including several of the world's major repositories for plant, animal and microbial genomes.
12	How does GO work? What does the gene product do? Where and when does it act? Why does it perform these activities? GO uses “GO term” to represent these concepts Each gene is associated (annotated) with multiple “GO terms” to describe its location and functions The information is stored in the GO database
13	The GO project (I) The GO project has developed three structured controlled vocabularies (ontologies) that describe gene products in terms of their associated biological processes, cellular components and molecular functions in a species-independent manner. There are three separate aspects to this effort: development and maintenance of the ontologies annotation of gene products, which entails making associations between the ontologies and the genes and gene products in the collaborating databases development of tools that facilitate the creation, maintenance and use of ontologies. The use of GO terms by collaborating databases facilitates uniform queries across them.
14	The Gene Ontology The Gene Ontology project provides an ontology of defined terms representing gene product properties. The ontology covers three domains pertinent to the functioning of integrated living units: cells, tissues, organs, and organisms. cellular component: the parts of a cell or its extracellular environment molecular function: the elemental activities of a gene product at the molecular level, such as binding or catalysis biological process: operations or sets of molecular events with a defined beginning and end
15	Example: GO terms for cytochrome c The gene product “cytochrome c” can be described by the following GO terms: molecular function: oxidoreductase activity biological process: oxidative phosphorylation and induction of cell death cellular component: mitochondrial matrix and mitochondrial inner membrane
16	The GO project (II) The controlled vocabularies are structured so that they can be queried at different levels. For example, you can use GO to find all the gene products in the mouse genome that are involved in signal transduction, or you can zoom in on all the receptor tyrosine kinases. This structure also allows annotators to assign properties to genes or gene products at different levels, depending on the depth of knowledge about that entity.
17	GO Structure
18	Structure of GO Terms The GO ontology is structured as a directed acyclic graph (DAC). Each term has defined relationships to one or more other terms in the same domain, and sometimes to other domains.
19	GO structure
20	GO structure
21	GO namespace GO terms are divided into three types: Cellular component : where and when does it act? Molecular function : what does the gene product do? Biological process : why does it perform these activities?
22	Cellular Component
23	Cellular Component
24	Cellular Component
25	Cellular Component Enzyme complexes in the component ontology refer to places, not activities.
26	Molecular Function & Biological Process A gene product may have several functions. A function term refers to a reaction or activity, not a gene product à How ? Sets of functions make up a biological process à Why ?
27	Molecular Function
28	Molecular Function insulin binding insulin receptor activity
29	Molecular Function drug transporter activity
30	Biological Process a commonly recognized series of events
31	Biological Process transcription
32	Biological Process regulation of gluconeogenesis
33	Biological Process limb development
34
35
36
37
38
39
40	Scope of GO Terms The GO vocabulary is designed to be species-neutral, and includes terms applicable to prokaryotes and eukaryotes, single and multicellular organisms.
41	Example 1 Using GO to identify all genes involved in a specific biological process.
42
43
44
45
46
47
48	Example 2 Using GO to classify genes differentially expressed from microarray study
49
50	Traditional Analysis After searching all information about these 100 genes, it is still difficult to know which biological processes are most significantly altered
51	Using GO Annotations But by using GO annotations, this work has already been done
52	Grouping Genes by Biological Process
53
54	How to spot biological functions embedded in a gene list?
55	DAVID Bioinformatics Resources DAVID web server : http://david.abcc.ncifcrf.gov/home.jsp
56	Construction of a DAVID Gene
57	Analytic tools/modules in DAVID
58	DAVID analytic modules
59	Gene List – Quality Control Reasonable number of genes ranging from hundreds to thousands (e.g., 100–2,000 genes), not extremely low or high. Most of the genes significantly pass the statistical threshold for selection (e.g., selecting genes by comparing gene expression between control and experimental cells with t-test statistics: fold changes ≥ 2 and P-values ≤0.05). A ‘good’ gene list should consistently contain more enriched biology than that of a random list in the same size range during analysis in DAVID.
60	Background List - Definition To decide the degree of enrichment, a certain background must be set up to be compared with the user’s gene list. For example, 10% of user’s genes are kinases versus 3% of genes in human genome (this is population background) are kinases. Thus, the conclusion is obvious in the particular example that the user’s study is highly related to kinase. However, 10% itself alone cannot provide such a conclusion without comparing it with the background information.
61	Background List – How to use A general guideline is to set up the reference background as the pool of genes that have a chance to be selected for the studied annotation category under the scope of users’ particular study Default background is the entire genome-wide genes of the species matching the user’s input IDs. Pre-built backgrounds, such as genes in Affymetrix chips and so on, are available for the user’s choice In principle, a larger gene background tends to give smaller P-values. As most of the high throughput studies are, or at least are close to, genome-wide scope, the default background is good for regular cases in general
62	Classification Stringency To control the behavior of DAVID Fuzzy clustering A general guideline is to choose higher stringency settings for tight, clean and smaller numbers of clusters; otherwise, lower for loose, broader and larger numbers of clusters Default setting is medium Five predefined levels from lowest to highest for user’s choices Users may want to play with different stringency for more satisfactory results
63	Enrichment Score - Definition To rank overall enrichment of gene groups. It is the geometric mean of all the enrichment P-values (EASE scores) for each annotation term associated with the gene members in the group. To emphasize that the geometric mean is a relative score instead of an absolute P-value, minus log transformation is applied on the average P-values. A higher score for a group indicates that the gene members in the group are involved in more important (enriched) terms in a given study; therefore, more attention should go to them
64	Fold Enrichment – How to use ? Enrichment score of 1.3 is equivalent to non-log scale 0.05. Fold enrichment 1.5 and above are suggested to be considered as interesting. Caution should be taken when big fold enrichments are obtained from a small number of genes (e.g., ≤3). This situation often happens to the terms with a few genes (more specific terms) or of smaller size (e.g.,<100) of user’s input gene list. In this case, the reliability is not as much as those fold enrichment scores obtained from larger numbers of genes
65	P-vlaue (EASE score) To examine the significance of gene–term enrichment with a modified Fisher’s exact test (EASE score). The smaller the P-values, the more significant they are Default cutoff is 0.1 Users could set different levels of cutoff through option panel on the top of result page. Owing to the complexity of biological data mining of this type, P-values are suggested to be treated as score systems, i.e., suggesting roles rather than decision-making roles. Users themselves should play critical roles in judging ‘are the results making sense or not for expected biology
66	Benjamini To globally correct enrichment P-values to control family-wide false discovery rate under certain rate (e.g., ≤0.05). It is one of the multiple testing correction techniques (Bonferroni, Benjamini and FDR) provided by DAVID More terms examined, more conservative the corrections are. As a result, all the P-values get larger It is great if the interesting terms have significant P-values after the corrections. But as the multiple testing correction techniques are known as conservative approaches, it could hurt the sensitivity of discovery if overemphasizing them.
67	% - Defintion Number of genes involved in given term is divided by the total number of user’s input genes, i.e., percentage of user’s input gene hitting a given term. For example, 10% of user’s genes hit ‘kinase activity It gives overall idea of gene distributions among the terms The higher percentage does not necessarily have a good EASE score because it also depends on the percentage of background genes
68	Data Interpretation Fold enrichment and EASE score should always be examined side by side. Terms with larger fold enrichments and smaller EASE score may be interesting.
69	Start analysis wizard
70	Submit gene list or use built-in demo gene lists
71	Select one of the DAVID Tools
72	Gene Name Batch Viewer
73	Gene Functional Classification
74	2D View of Gene Function Classification
75	Select annotation category and run Functional Annotation Chart
76	Select annotation category and run Functional Annotation Chart
77	Select annotation category and run Functional Clustering
78	Functional Table
79	DAVID Bioinformatics Resources DAVID web server : http://david.abcc.ncifcrf.gov/home.jsp