1. Microbial Genome Annotation
Nikos Kyrpides
DOE Joint Genome institute 1

2. Two main goals of genome analysis:
Evolutionary analysis
How does an organism compare to the rest?
Metabolic reconstruction
What can an organism do and how?

3. Overview of Annotation Steps
DNA
sequence
>Contig1
ataacaacacattagcggc
asacacacaacaggatatt
aggagagagagaaagttac
Gene Finding
Function Prediction
Identify Genes
(Proteins, RNAs)
Identify Regulatory elements
Identify Repeat elements
Blast
Clusters
(BBH, COGs, TIGRFam)
Motifs
(HMM, Pfam, InterPro)
Automatic
Manual
Gene QC
Gene Context
(Fusions, Operons, Regulons)
Missing Genes

4. 1. Finding the genes in microbial genomes
Introduction
2. Tools out there
3. Basic principles behind tools
4. Known problems of the tools: why you may need manual curation

5. Finding the genes in microbial genomes
features
Sequence features in prokaryotic genomes:
stable RNA-coding genes (rRNAs, tRNAs, RNA component of RNaseP, tmRNA)
protein-coding genes (CDSs)
transcriptional features (mRNAs, operons, promoters, terminators, protein-binding sites, DNA bends)
translational features (RBS, regulatory antisense RNAs, mRNA secondary structures, translational recoding and programmed frameshifts, inteins)
pseudogenes (tRNA and protein-coding genes) …
Fairly big project in terms of number of institutes and people involved

6. Tools out there: finding protein-coding genes (not ORFs!)
Reading frames: translations of the nucleotide sequence with an offset of 0, 1 and 2 nucleotides (three possible translations in each direction)
Open reading frame (ORF): reading frame between a start and stop codon

7. Finding features in microbial genomes
Well-annotated bacterial genome in Artemis genome viewer:

8. Finding the genes in microbial genomes
Introduction
2. Tools out there
3. Basic principles behind tools
4. Known problems of the tools: why you may need manual curation

9. Tools out there: servers for microbial genome annotation - I
IMG-ER
IMG-ER submission page:
RAST
JCVI Annotation Service
Output:
stable RNA-encoding genes,
CDSs,
functional annotations
output in GenBank format
Output:
rRNAs and tRNAs,
CDSs,
functional annotations
output in several formats
Output:
CDSs, stable RNAs?
functional annotations
format?

10. Tools out there: servers for microbial genome annotation - II
AMIGENE
RefSeq
EasyGene
Output:
CDSs,
output in gff format
Output:
CDSs,
output in tbl format
Output:
CDSs,
size restriction <1Mb

11. Major difference: viewer vs editor?
Tools out there: genome browsers for manual annotation of microbial genomes
Artemis
Manatee
Argo
Major difference: viewer vs editor?
Windows and Linux versions;
works with files in many formats, annotated by any pipeline
Linux versions only;
genome needs to be annotated by the JCVI Annotation Service
Windows and Linux;
works with files in many formats

12. Tools out there: tools for finding stable (“non-coding”) RNAs - I
Large structural RNAs (23S and 16S rRNAs)
RNAmmer
Small structural RNAs (5S rRNA, tRNAs, tmRNA, RNaseP RNA component)
Rfam database, INFERNAL search tool
ARAGORN
tRNAScan-SE
Web service: sequence search is limited to 2 kb
Web service: sequence search is limited to 15 kb, finds tRNAs and tmRNAs only
Web service: sequence search is limited to 5 Mb, finds tRNAs only

13. Tools out there: tools for finding “non-coding” RNAs - II
Short regulatory RNAs
Rfam database, INFERNAL search tool
Other (less popular) tools:
Pipeline for discovering cis-regulatory ncRNA motifs:
RNAz
Web service: sequence search is limited to 2 kb;
Provides list of pre-calculated RNAs for publicly available genomes

14. Tools out there: most popular CDS-finding tools
CRITICA
Glimmer family (Glimmer2, Glimmer3, RBS finder)
GeneMark family (GeneMark-hmm, GeneMarkS)
EasyGene
AMIGENE
PRODIGAL (default JGI gene finder)
Combinations and variations of the above
RAST (Glimmer2 + pre- and post-processing)

15. Basic principles: finding CDSs using evidence-based vs ab initio algorithms
Two major approaches to prediction of protein-coding genes:
“evidence-based” (ORFs with translations homologous to the known proteins are CDSs)
Advantages: finds “unusual” genes (e. g. horizontally transferred); relatively low rate of false positive predictions
Limitations: cannot find “unique” genes; low sensitivity on short genes; prone to propagation of false positive results of ab initio annotation tools
ab initio (ORFs with nucleotide composition similar to CDSs are also CDSs)
Advantages: finds “unique” genes; high sensitivity
Limitations: often misses “unusual” genes; high rate of false positives

16. Finding the genes in microbial genomes
Introduction
Tools out there
Basic principles behind tools
Known problems of the tools: why you may need manual curation

17. Known problems: CDSs
Short CDSs: many are missed, others are overpredicted
short ribosomal proteins (30-40 aa long) are often missed
short proteins in the promoter region are often overpredicted
N-terminal sequences are often inaccurate (many features of the sequence around start codon are not accounted for)
Glimmer2.0 is calling genes longer than they should be
GeneMark, Glimmer3.0 mostly call genes shorter
Pseudogenes and sequencing errors (artificial frameshift)
all tools are looking for ORFs (needs valid start and stop codons)
“unique” genes are often predicted on the opposite strand of a pseudogene or a gene with a sequencing error
Proteins with unusual translational features (recoding, programmed frameshifts)
these genes are often mistaken for pseudogenes
see pseudogenes

18. Known problems: CDSs
Lack of Standards

19. Finding unique genes Obligate parasite of horses
Causes human disease in tropical areas (melioidosis)

20. 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

22. Pati A. et al, Nature Methods June 2010
GenePRIMP
GenePRIMP
Gene Prediction Improvement Pipeline
GenePRIMP is a pipeline that consists of a series of
computational units that identify erroneous gene
calls and missed genes and correct a subset of the
identified defective features.
APPLICATIONS
Identify gene prediction anomalies
Benchmark the quality of gene prediction algorithms
Benchmark the quality of combination / coverage of sequencing platforms
Improve the sequence quality
Pati A. et al, Nature Methods June 2010

23. GenePRIMP steps

24. Find missing genes
Intergenic regions identify missed ORFs …

25. … and wrong ORFs or2654 is unique and hides a real CDS which
is acyl carrier protein

26. Everything looks perfect in this area of Nitrobacter winogradskyi, but …

27. … hides a real ORF

28. Guinness Book of protein-coding genes
The longest human gene is 2,220,223 nucleotides long. It has 79 exons, with a total of only 11,058 nucleotides, which specify the sequence of the 3,685 amino acids and codes for a protein dystrophin. It is part of a protein complex located in the cell membrane, which transfers the force generated by the actin-myosin structure inside the muscle fiber to the entire fiber.
The smallest human gene is 252 nucleotides long, it specifies a polypeptide of 67 amino acids and codes for an insulin-like growth factor II.
The longest bacterial gene is 110,418 nucleotides long, which specify the sequence of 36,805 amino acids. Its function is unknown, most likely a surface protein.
The smallest bacterial gene is 54 nucleotides long, it specifies a polypeptide of 17 amino acids and codes for a regulatory protein in cyanobacteria

29. False positives Genome name CDSs with no hits < 100 aa
% with tBLASTn hit
% tBLASTn hits with frameshifts/stop codons
Prochlorococcus AS9601 18
88.9
68.8
Prochlorococcus MIT 9211 62
40.3 80
Prochlorococcus MIT 9215 24
58.3
64.2
Prochlorococus MIT 9301 12 75
66.7
Prochlorococcus MIT9303
501 83
61.8
Prochlorococcus MIT 9313 35
8.6
Prochlrococcus MIT 9515 32
81.3 50
Prochlorococcus NATL1A
209
95.2
48.2
Prochlorococcus CCMP1375 34
82.4
Synechococcus PCC 7942 39
Synechococcus CC9311
313
11.5
83.3
Synethococcus CC9605
38.6
Synechococcus CC9902 21
57.1
100
Synechococcus JA-2-3Ba
176
26.7
85.1
Synechococcus JA-3-3Ab
142
35.2 92
Synechococcus PCC 7002 93
17.2
56.3
Synechococcus RCC307
184
10.3
68.4S
Synechococcus WH 7803
18.8
Synechococcus WH 8102
38.4
46.7

30. 2. Finding the functions in microbial genomes
Introduction
2. Tools out there
3. Known problems

31. what is function?
cobalamin biosynthetic enzyme, cobalt-precorrin-4 methyltransferase (CbiF)
molecular/enzymatic (methyltransferase)
Reaction (methylation)
Substrate (cobalt-precorrin-4)
Ligand (S-adenosyl-L-methionine)
metabolic (cobalamin biosynthesis)
physiological (maintenance of healthy nerve and red blood cells, through B12).

32. Functional characterization

33. Computational approaches to Functional characterization

34. Sequence Homology
Two sequences are homologous, if there existed a molecule in the past that is ancestral to both of the sequences.
Types of Homology:
Orthology: bifurcation in molecular tree reflects speciation
Paralogy: bifurcation in molecular tree reflects gene duplication

35. Homology & analogy
The term homology is confounded & abused in the literature!
sequences are homologous if they’re related by divergence from a common ancestor
analogy relates to the acquisition of common features from unrelated ancestors via convergent evolution
e.g., b-barrels occur in soluble serine proteases & integral membrane porins; chymotrypsin & subtilisin share groups of catalytic residues, with near identical spatial geometries, but no other similarities
Homology is not a measure of similarity & is not quantifiable
it is an absolute statement that sequences have a divergent rather than a convergent relationship
the phrases "the level of homology is high" or "the sequences show 50% homology", or any like them, are strictly meaningless!
Homology:bat’s wing and human’s hand
Analogy: bat’s wing and butterfly’s wing

36. Function prediction Function transfer by homology Homology
implies a common evolutionary origin.
not retention of similarity in any of their properties.
Homology ≠ similarity of function.
Punta & Ofran. PLOS Comp Biol. 2008

37. Dos and Don’ts Type Don’t Do Homology Same function
Probability for same function
Orthology
Paralogy
Sequence similarity
High sequence similarity
Same sequence

38. Application areas of analysis tools
The scale indicates % identity between aligned sequences
Alignment of 2 random seqs can produce ~20% identity
less than 20% does not constitute a significant alignment
around this threshold is the Twilight Zone, where alignments may appear plausible to the eye, but can’t be proved by conventional methods

39. Finding the functions in microbial genomes
Introduction
2. Tools out there
3. Known problems

40. Function prediction
Similarity searches (BLAST).
Domain identification(Pfam).
Small sequence identification(PROSITE).

41. What if nothing is similar ?
Subcellular localization
Gene context
Structure
Prediction of binding residues (DISIS, bindN)
Cytoplasm
S ~ S
Periplasm

42. Annotation should make sense
Model pathway
Substrate A
Substrate B
Substrate C
Substrate D
Enzyme 1
Enzyme 3
Enzyme 2
Genome annotation
Enzyme 1
Enzyme 3
Enzyme 2 ?

43. Annotation should make sense

44. Databases Databases used for the analysis of biological molecules.
Databases contain information organized in a way that allows users/researchers to retrieve and exploit it.
Why bother?
Store information.
Organize data.
Predict features (genes, functions ...).
Predict the functional role of a feature (annotation).
Understand relationships (metabolic reconstruction).

45. Primary nucleotide databases
EMBL/GenBank/DDBJ (
Archive containing all sequences from:
genome projects
sequencing centers
individual scientists
patent offices
The sequences are exchanged between the three centers on a daily basis.
Database is doubling every 10 months.
Sequences from >140,000 different species.
1400 new species added every month.
Database name nt / nr
Year Base pairs Sequences
,575,745,176 40,604,319
,037,734,462 52,016,762
,019,290,705 64,893,747
,874,179,730 80,388,382
,116,431,942 98,868,465

46. Primary protein sequence databases
Contain coding sequences derived from the translation of nucleotide sequences
GenBank
Valid translations (CDS) from nt GenBank entries.
UniProtKB/TrEMBL (1996)
Automatic CDS translations from EMBL.
TrEMBL Release 40.3 (26-May-2009) contains 7,916,844 entries.

47. RefSeq Curated transcripts and proteins. reviewed by NCBI staff.
Model transcripts and proteins.
generated by computer algorithms.
Assembled Genomic Regions (contigs).
Chromosome records.

48. Classification databases
Groups (families/clusters) of proteins based on…
Overall sequence similarity.
Local sequence similarity.
Presence / absence of specific features.
Structural similarity.
...
These groups contain proteins with similar properties.
Specific function, enzymatic activity.
Broad function.
Evolutionary relationship. …

49. Overall sequence similarity

50. Clusters of orthologous groups (COGs)
COGs were delineated by comparing protein sequences encoded in 43 complete genomes representing 30 major phylogenetic lineages.
Each Cluster has representatives of at least 3 lineages
A function (specific or broad) has been assigned to each COG.

51. How it works COG1 COG2 Reciprocal best hit Bidirectional best hit
Blast best hit
Unidirectional best hit
COG1
COG2

52. Profiles & Pfam
A method for classifying proteins into groups exploits region similarities, which contain valuable information (domains/profiles).
These domains/profiles can be used to detect distant relationships, where only few residues are conserved.

53. Regions similarity

54. Pfam HMMs of protein alignments (local) for domains,
HMMs of protein alignments
(local) for domains,
or global (cover whole protein)

55. PROSITE http://au.expasy.org/prosite/
R-Y-x-[DT]-W-x-[LIVM]-[ST]-T-P-[LIVM](3)

56. KEGG orthology

57. Composite pattern databases
To simplify sequence analysis, the family databases are being integrated to create a unified annotation resource – InterPro
Release 32.0 (Apr 11) contains entries
Central annotation resource, with pointers to its satellite dbs

58. * It is up to the user to decide if the annotation is correct *

59. KEGG
Contains information about biochemical pathways, and protein interactions.

60. Summary
We have main archives (Genbank), and currated databases (Refseq, SwissProt), and protein classification database (COG, Pfam).
This is the tip of the iceberg of databases.
They help predict the function, or the network of functions.
Systems that integrate the information from several databases, visualize and allow handling of data in an intuitive way are required

61. Functional annotation in IMG
Automated protein product assignment pipeline
Functional context in IMG
KEGG Pathways, Modules, KEGG Orthology
MetaCyc Pathways
IMG Pathways
No longer maintained:
TIGR Role Categories
TIGR Genome Properties
COG Functional Categories

62. Lack of Standards