1. The Integrated Microbial Genome (IMG) systems
Nikos Kyrpides 1

2. Data analysis
Data Integration
Comparative Analysis

3. Data management system for comparative analysis of biological data
What is the Matrix?
Data management system for comparative analysis of biological data
IMG
Genes
Genomes
Functions
Metadata
Clusters
SNPs
Proteomics
Regulons
Transcriptomes I M G

4. Integrated Microbial Genomes (IMG) [It’s easier to analyze 1000 genomes than a single one]
What is IMG:
IMG is a data management system for comparative analysis and annotation of all publicly available genomes from three domains of life in a uniquely integrated context.
Mission:
To become the Home of Microbial Genome and Metagenome Analysis
Background:
 Launched on March 2005
 3 Releases/Year
 >5,000 unique visitors per month
 >300 citations
Current Status:
8939 Genomes
24 Million Genes
Bacteria: 3930
Archaea:
Eukarya: 177
Plasmids: 1205
Viruses:
USERS CAN
Search data
Browse data
Compare data
Export data
Gfragments:654

5. http://img.jgi.doe.gov/ USERS CAN Search data Browse data Compare data
Export data
USERS CAN
Submit data
Annotate data

6. Data Model Abstraction Example: IMG Operations
Genes present in G1
and absent from G2, G3, G4 and G5
G1 G2 G3 G4 G5 g3 g2 g1
Gene occurrence profile across genomes
Genes
Gene occurrence profiles across pathways
Genomes
Pathways shared by genomes
Perhaps you can mention that the dimensional modeling approach has a positive impact in data exploration. 1 and 2 are examples of slice and dice and the result is data reduction and focus on relevant to the question data set.
Functions/ Pathways

7. IMG Data Integration Genes Genomes Functions 24.2M 8939 1.1M
COG GO
Pfam
TIGRfam
InterPro
KEGG
BioCyc
SEED
Protein product
MyIMG
IMG Terms
IMG Pathways
IMG Networks
Groupings
Phylogenetic
Phenotypic
Ecotypic
Disease
Geographical
Isolation
RNAs, Proteins
Sequence Clusters
Positional clusters
Regulatory clusters
Fusions
Operons
Expression
Genes
24.2M
Genomes
Functions
8939
1.1M

8. IMG Toolkit Chromosome Map Function Profile Gene Synteny Abundance
Profiles
Functional
Categories
Projects
IMG Pathway
Metadata
Search
Phylogenetic
Genome
Clustering
Compare
Annotations
KEGG
Maps
Distribution
Chromosomal
Artemis
VISTA
Recruitment
Plot
Fragment

9. Challenges and Opportunities
Annotations
Annotations
Quality
Metadata
Genes
Functions
Data
Analysis
New data types and tools
Integration
# genes and genomes
Scaling

10. Metadata Curation Metadata Types Organism Information
K. Liolios
Metadata Types
Organism Information
Genome Project Information
Sequencing Information
Environmental Metadata
Host Metadata
Organism Metadata

11. Metagenome Classification
Genomes vs Metagenomes

12. The negative example or why we need high quality data
If we run the same 3-click query, we find that Burkholderia mallei (the smaller one) still has 548 genes with no match in Burkholderia pseudomallei.
Are these genes important for its lifestyle?
No, they are not. 89% of these genes are actually due to the difference in gene prediction algorithms used by different sequencing centers. That’s why in IMG we have implemented a validation and correction procedures. We have started with JGI genomes and will further extend them to all public genomes, so that users will be able to see only the real differences between organisms and not the false positives.
Phylogenetic profiler finds 548 unique genes in B. mallei
However, 497 of them in fact exist in B. pseudomallei, but they have not been called as real genes.
The difference in gene models reveals 89.2% error rate in unique genes

13. Gene Prediction - Standards
Re-Annotation workshops
- January 19-20, JGI
- February 29,
Ivanova
Usage of
Reference Genomes
Annotation of isolate genomes, single cells and metagenomes
Computation of Gene cassettes
Computation of Pangenomes
Problems with current Public
Reference Genomes
lack of provenance for the predicted features
presence of artificial (non-biological) variation between the genomes including variation in gene content and variation within protein families

14. Constant Benchmarking
Evaluation of Annotation Quality with constantly changing:
K. Mavrommatis
Sequencing technologies
Read lengths
Gene calling methods
Similarity methods
Clustering methods
Functional annotation methods

15. Blat & Uclust vs Blast

16. Program Informatics Challenges and Opportunities
Annotations
Quality
Data
Analysis
New data types and tools
Integration
# genes and genomes
# genes and genomes
Scaling

17. Why annotate unassembled reads?
Kansas soil
Total size
102,722,384 (2x150) reads
Assembled contigs
1,375,950 contigs
Assembled (reported by the CLC workbench assembler)
38,094,033 reads
5060 different pfams
Assembled reads
Mapped (by bwa)
11,778,925 reads
Genes called on unassembled reads
64,737,444 genes
7481 different pfams
8,373,641 (12%) genes
Similar to genes on contigs1
Genes with similarity to isolate genomes
40,778,854 genes
Additional information about functions and phylogeny
Assembled only
More accurate statistics based on unassembled + assembled
Unassembled + assembled +
real metagenome

18. Annotating unassembled Illumina data
SEPTEMBER 2011
Samples
937
DNA (bps)
84 B
Private Genes
188 M
Average Illumina Metagenome
Sequences
673,374,734
Bases
64,545,005,513
Genes
667,966,495
Genes with COGs
14%
Genes with Pfam
8.8%
Genes with KO 6%
MAY 2012
Samples
997
DNA (bps)
608 B
Private Genes
6.03 B

19. Where do we go from here?

20. DELUGE
AVALANCHE
FLOOD
TSUNAMI
OPPORTUNITY

21. Program Informatics Challenges and Opportunities
Annotations / Publications
Quality
Data
Analysis
New data types and tools
New data types and tools
Integration
# genes and genomes
Scaling

22. Challenges and Opportunities
Gene Clustering
Metagenome Classification
Data
Analysis

23. MGM Workshop Attendees
Europe:
Belgium
Czech Rep
Denmark
Estonia
Finland
France
Germany
Greece
Ireland
Italy
Hungary
Netherlands 4
Norway
Russia
Portugal
Poland
Spain
Sweden
Switzerland 1
UK 10
Asia:
China
Hong Kong
India
Israel
Japan
Korea
Malaysia
Philipines
Saudi Arabia 4
Singapore
Taiwan
Thailand
Turkey
North America: 356
Canada
Mexico
USA
South America: 21
Argentina
Brazil
Chile
Colombia
Ecuador
Peru
Uruguay
Africa:
Algeria
Egypt
Ethiopia
Oceania:
Australia
New Zeeland 2
545 /48 Countries
April 20, 2012

24. Unique properties of IMG
Largest metadata integration from GOLD
Largest integration of Genes (> 4 Billion)
Clustering of all metagenomic genes
IMG
QC of all gene predictions from isolate genomes
Metagenome classification scheme
Large array of function & pathway analysis tools