pre-sig genome annotation database operations

Pre-SIG Genome AnnotationDatabase Operations

Suzanna Lewis

FlyBase/Berkeley Drosophila Genome ProjectGene Ontology Consortium

Having it all

Complete: every occurrence is found

Precise: every occurrence is accurate

Comprehensive: all types of features

Richly described: biological functional data

Contradictions and Complications

Assembly errors Missed gene merges Missed gene splits Complex adjustments Dicistronic genes Overlaps and intersections

Assembly dependencies

Missed merges

Missed Splits

Splerges

Dicistronic Genes

Shared 5’ UTR

Shared 3’ - 5’ UTRs

Having it all





deleted: 41new: 179merges: 31splits: 26reinstated: 32

Broad Institute

TIGR

JGI

Baylor College of Medicine

Washington University

FlyBase

Ensembl

GMOD contributors meeting March 2004

The Essentials

Visualization and manual editors

The Essentials

Visualization and manual editors Combiners

The Essentials

Visualization and manual editors Combiners Full-length cDNA sequences

The Essentials

Visualization and manual editors Combiners Full-length cDNA sequences High-quality assemblies

The Essentials

Visualization and manual editors Combiners Full-length cDNA sequences High-quality assemblies Annotation standards and

verification

The Essentials


verification Evidence tracking and versioning

The Essentials


verification Evidence tracking and versioning Open source software components

and standards are critical to long term success

Annotation verification

Community input On-line error reporting Curation of the literature

Confirmation by comparison to cDNA sequences

SWISSPROT Comparison

Perfect match 100% identity over 100% of lengths

Single AA substitutions 99% identity over 100% of lengths

The above account for 75% of all genes with a SWISSPROT cognate (2,771 out of 3,687).

SWISSPROT Comparison

Significant mismatch spans of 40 residues or 20% peptide

length, with at least 97% sequence identity

No match poor or empty matches

These remaining differences were due to lingering annotation errors or errors in the reported DNA sequence from SWISSPROT

Analysis of 8687 cDNAs (full inserts)

Having it all





Annotation of all types of features

Protein-coding 13,410

tRNA 291

microRNA 23

snRNA 32

snoRNA 29

Pseudogenes 19

Non-coding RNA 36

Transposons 1,572

Promoters (thousands)

TSS (thousands)

P element Insertions (thousands)

Total 15,412

Having it all





How to find what you need

FlyBase MGISGD

Cappuccino BNI1 Formin 2

By name?By database ID?

Actin binding Actin binding Actin binding

FBgn0000256

S0005215 MGI:1859252

By function?

But in 1998 there was a problem…

None of the organism databases used standard terminology to describe biological function.

For example

It will be difficult for you—and even harder for a

computer—to find functionally equivalent gene

products.

translation Proteinsynthesis

You want all gene products that are involved in bacterial protein synthesis, But the sequences are significantly different from those in humans.

How to best describe biology?

Natural language Highly expressive Ambiguous in meaning Hard to compute on

Structured

representation Limited in expressivity Precise May be computed on

We needed to find a middle ground,

that supports and enables both.

The aims of GO

1.To develop comprehensive shared vocabularies.

2.Use the vocabularies to describe the gene products held in different databases.

3.To provide access to the vocabularies, the annotations, and associated data.

4.To provide software tools to assist biological researchers.

The early key decisions

The vocabulary itself requires a serious and ongoing effort.

Carefully define every concept Initially keep things as simple as

possible and only use a minimally sufficient data representation.

Focus initially on molecular aspects that are shared between many organisms.

A sequence is not equal to a gene

Physically a gene is composed of sequences.

DNA, RNA, and protein Different strains, ESTs, cDNAs,

alleles… A fully characterized gene has

multiple sequence references

GO is NOT a gene nomenclature system

Communities decide upon the ‘official’ gene name or symbol and their community databases maintain these data.

Sequence repositories (I.e. Genbank/EMBL/DDBJ/SwissProt) provide sequence identifiers and protein names

Proteins may be named differently than genes e.g. HUGO and UniProt IDs

GO encompasses descriptions for

all functional molecular entities A gene product may be either a

functional RNA or a protein Protein tRNA miRNA snRNA rRNA …

The breakdown of work

Task 1 Building the ontology: a computable

description of the biological world

Task 2 Describing your gene—annotation

Protein structure Phenotype Expression data Function, process, localization…

Vocabulary and relationships

Look up concept to accurately

express biology Your gene product

Refer to representative

sequences

Gene nomenclaturedecisions

SequenceDB

Choose approved name and synonyms

Collect what is known from the literature

GO databases: distributed and centralized

Support cross-database queries By having a mutual understanding of the

definition and meaning of any word used to describe a gene product

Provide database access to a common repository of annotations By submitting a summary of gene products

that have been annotated

If we build it…

What is a term? Definition of term concepts How to represent and manage the

concepts Biological scope Annotation

What is a ‘term’?

Must have a stable ID

May have synonyms

Have relationships to other terms

Can be made obsolete

Can be split or merged

Must have a definition

Definitions

Purpose is to remove ambiguity of interpretation and alternate meanings

All definitions are supported by cross-references to the source(s)

Annotating gene products Expert curation accepted from any group

that can provide an ID and evidence Each annotation must be supported by

evidence, including a cross-reference http://www.geneontology.org/GO.evidence.html

Gene can be annotated with multiple terms Annotation is at finest possible granularity Guidelines

http://www.geneontology.org/doc/GO.usage.html

GO functional analysis

Sequence similarity Literature harvesting Motif analysis Expression studies Interaction studies

Current work

Low coverage genome sequencing Multiple species genome

comparisons SO—the Sequence Ontology

e.g. What is a pseudogene?

Human Sequence similar to known protein but

contains frameshift(s) and/or stop codons which disrupts the ORF.

Neisseria A gene that is inactive - but may be

activated by translocation (e.g. by gene conversion) to a new chromosome site.

- note some would call such a gene a “cassette” in yeast.

SO is useful if you want to:

Annotate sequence using consistent terminology for the same features across genomes.

Enable practical querying and comparisons between sequence databases.

Describe and propagate features at all levels of the sequence from genomic to mature protein.

Thank You to…www.flybase.org

Curators-Berkeley Sima Misra, Josh

Kaminker, Simon Prochnik, Chris Smith, Jon Tupy

Curators-Harvard Lynn Crosby, Bev

Matthews, Kathy Campbell, Pavel Hradecky, Yanmei Huang, Leyla Bayraktaroglu

Curators-Cambridge Gillian Millburn, Rachel

Drysdale, Chihiro Yamada

Curators-SWISSPROT Eleanor Whitfield

Software-Berkeley Chris Mungall, Ben Berman,

Joe Carlson, Mark Gibson, Nomi Harris, George Hartzell, Brad Marshall, John Richter, ShengQiang Shu

Software-Harvard David Emmert

Software-Cambridge Aubrey de Grey

Software-Ensembl Michelle Clamp, Vivek Iyer,

Steve Searle

and Thanks to…

www.geneontology.org Christopher Mungall

John Day-Richter

Brad Marshall Karen Eilbeck Mark Yandell George Hartzell

David Hill Joel Richardson GO Curators Michael Ashburner Judith Blake J. Michael Cherry

We want and depend on you!

Corrections to the peptides Functional annotation Corrections and additions to GO

pre-sig genome annotation database operations

Documents