1. Sequence, SNP and Mutation Databases
I’m mesut, phd student in genetic epidemiology studying the effects of consanguinity on human disorders - disorder called PCD
I’m here to talk to you about databases
We’ll have two exercises at the lecture which will be very useful to you and for this reason we have Tom and Denis with us also
I hated lectures before the coffee break
Probably the most boring but also the most useful lecture – there are hundreds of databases out there so we can’t go through all of them
However not all of them are applicable to you anyway – so within the next 45 minutes I’ll talk to you about the ones I found the most useful
*In front of very important databases – especially Ensembl and GeneCards
Professor asked me to explain
Feel free to stop me at anytime but there is a lot to go through so unless it is urgent please leave it to the end – also feel free to send me s
Mesut Erzurumluoglu

2. Before we start Very important lecture Hundreds of databases!
Database knowledge is a must for analysis
Hundreds of databases!
Pay special attention to:
Ensembl Genome Browser
dbSNP
GeneCards
Exercise and Q&A time at the end

3. Two lines of data Mapping of locations – Annotation
Genomic landmarks
e.g. Genes (exon, intron, splice site), binding sites
Known associations
e.g. HBB gene and sickle cell disease
rs SNP and BMI
Recording of variation – Summary stats
Frequencies, counts
Correlations (e.g. SNP-SNP)
So there are two lines of data in these databases: one to with the mapping of locations. So what I mean by this is it would make no sense to just dump the sequence some bit of DNA and dump it into a database. You’d have to make it clear which species is it from, is it from a healthy individual or suffering from some disorder. Also you would want to know where the sequence lies in the genome, whether the sequence is coding, is it functional, what does it do? If mutated, does it cause a disease. Finally the second line of data is the recording of variation. So if you have a variation you’re working on, how many times has it been seen in other individuals.

4. Why is it important? Mapping of locations Recording of variation
Designing experiments
e.g. Primers, candidate genes
Location of variants
e.g. genes within region, exon, intergenic
Interpretation of results
e.g. LD, consequence
Recording of variation
What’s out there from other projects?
Standardisation
There is a standard – so we all know we’re talking about the same loci when we say chr 11, position 1000

5. Terminology Single nucleotide polymorphism (SNP)
Minor allele frequency (MAF)
Difference between genotyping and sequencing
Linkage disequilibrium (LD)
Will be taught on next short-course
Slide in appendix
Genetic association study
Also taught on next short-course
In order to benefit from this lecture and databases completely you’d have to understand these terms especially. I’ll talk about the first three but the other two will require a lot of time to explain thus I’ll let go of them. However I did make two slides available at the end for those who are not going to the next short course.

6. SNP Affects single nucleotide Common ones (>1%)
Mostly bi-morphic (e.g. A>C)
Minor allele
Second most common one
Minor allele frequency
Frequency of minor allele
SNV or SNP?
Scottish national party
99.5% similarity
0.5% * 6billion = 30 million mismatches = 3 million SNVs; others are indels, CNVs etc.
More in Africans than Europeans
Why sequence the whole genome, which is very expensive, when you can just pinpoint these loci which are polymorphic and just genotype them
So large scale projects were initiated to find these polymorphic sites and then they designed micro-arrays to just genotype these loci. Of course you miss out on the unique and the rare mutations but still captured most of the variation this way in cheap and quick fashion.
Common disease common variant hypothesis – since 1996

7. Data formats FASTA – raw sequencing data
Can be nucleotide or protein
SAM/BAM – sequence alignment format
VCF – only variations from reference
Nucleotide
Can contain SNPs, insertion and/or deletions (aka indels), microsatellites
PED/MAP – Plink default format
Incorporate familial and phenotypic data to genetic data
Slide in appendix
VEP – Ensembl’s variation annotation format

8. FASTA (and FASTQ) Quality scores Sample ID
DNA Sequence (or amino acid)
Very large file sizes
! Lowest and ~ highest
Numbers (0 to 9): low-mid
Capital letters (A to Z): mid-high
Small letters (a to z): high-higher

9. VCF Small file size and widely used
Very useful for Mendelian disease studies
Software: Vcftools
Reference human genome: GRCh37 (release 74) -

10. Raw sequence databases
Ensembl (Europe), NCBI (USA)
DNA
NC_ (complete genome)
NG_ (genomic region)
mRNA
NM_ (mRNA)
NR_ (transcript)
Protein
NP_ (protein)
Also see RefSeq (NCBI) slide in appendix
Ensembl is European, NCBI is USA based
Ensembl is neater but NCBI has more data
XM_,XR_,XP_ are not curated

11. *Ensembl www.ensembl.org/info/data/ftp/index.html DNA
Useful for mapping reads, designing primers
cDNA = mRNA (transcripts)
CDS = all exons in gene
VCF
HapMap, 1000 Genomes, Venter, Watson
VEP
66 species (as of 26/02/14)

12. Genome Browsers Ensembl Genome Browser UCSC Genome Browser
View region
Extract region
Filter variants in region
Prediction
MAF
Conservation
GERP scores
UCSC Genome Browser
1000 Genomes Browser
VEGA Genome Browser

13. *Ensembl Genome Browser (EGB)
Browse many genomes (>70 vertebrates)
Example: BRCA2
Start and end
Sequence (exons, introns)
Transcripts
Orthologues
Sequence homology in other species
Indicative of similar function
Easy sequence extraction:
Other EGBs
EnsemblPlants, EnsemblBacteria, EnsemblFungi, EnsemblMetazoa, EnsemblProtists

14. EGB – Conservation (GERP)
Location
GERP elements

15. SNP databases Ensembl (and NCBI) dbSNP (largest database for SNPs)
1000 Genomes
1092 whole genome sequencing
African, European, Far East
HapMap projects
Genotyping 270 individuals from 4 populations
Exome Variant Server (EVS)
6503 individuals’ whole exomes
American European and American African
SNP and Phenotype associations
GWAS Catalogue
Slide in appendix

16. *dbSNP http://ncbi.nlm.nih.gov/SNP Search, annotate and submit SNPs
Apply filters
e.g. human, cited, clinical
>70 million SNPs just for humans
dbSNP handbook

17. From GWAS catalogue One of the most famous figures out there
Y chromosome
From GWAS catalogue

18. Clinical databases OMIM HGMD mutDB Mitomap (mitochondrial DNA)
Online Mendelian Inheritance in Man
Database of disease-linked genes and associated phenotypes
Links to Entrez, GDB and other databases
HGMD
Database of sequences and phenotypes of disease-causing mutations
Used to train mutation effect prediction algorithms (e.g. FATHMM)
Slide in appendix
mutDB
Mitomap (mitochondrial DNA)

19. *OMIM http://omim.org/ Online Mendelian Inheritance in Man
First address for Mendelian disorders
Disease > gene and phenotype
Gene > disease and phenotype
Phenotypes > disease and genes
Example search:
Autosomal recessive intellectual disability
+autosomal +recessive +intellectual +disability

20. Protein databases Uniprot (Swiss-Prot + TrEMBL)
PDB – Protein data bank
Slide in appendix
Pfam – Protein families
STRING – Protein-Protein Interactions
HMPD (mitochondrial DNA)

21. Uniprot http://www.uniprot.org/ Swiss-Prot is manually curated
TrEMBL is automatically curated
Download protein sequence
Protein sequence BLAST
Align
Conserved regions and/or residues
View AA properties
Try example:
Gene: DNALI1
Retrieve FASTA sequence, Blast and align (3 species)

22. Very useful integrative databases
GeneCards
Integrated resource of information on human genes and their products
Major emphasis on human disease
Links to many kinds of biomedical information
Sequence databases
OMIM, HGMD, MDB
Doctors’ Guide to the Internet
Ensembl (and NCBI) fits in this category

23. *Genecards http://www.genecards.org/
Graphical view of many things about your gene
Links to Ensembl, OMIM and Literature
Example:
DNALI1
Entrez Gene Summary
Associated disorders
Orthologues
STRING predictions

24. Google is your friend! PubMed – Bibliographic database Animal models
Zfin – Zebrafish knockouts
International Mouse Phenotyping Consortium
New ones coming out all the time!
New cohorts and studies
UK10K project
Human Microbiome project
ARIES - Epigenetics

25. *PubMed http://www.ncbi.nlm.nih.gov/pubmed/
“>23 million citations for biomedical literature from MEDLINE, life science journals, and online books”
Simple to use
Searching
Citation manager facility
PubMed help book
And finally being able to search the literature efficiently and accessing the right ones is very important; and I’ve found PubMed a very reliable and easy to use database for bibliographic sources

26. Exercise 1 - SNPs Find where rs9939609 is located
Is it in an exon or an intron?
Minor allele? Global MAF?
Which gene(s) are close by?
Associated with any disorders?
Which population(s) is the minor allele most frequent in?
How many other known human SNPs in this gene?

27. Answers Chr 16 at position 53,820,527 Intron A, 0.355 FTO
BMI, Type 2 diabetes, Menarche
Luhya in Webuye, Kenya (0.617)
dbSNP, scroll to bottom
8099 (as of 21/02/14)
Search for FTO in dbSNP and filter for H.sapiens

28. Exercise 2 - Genes
Find the start and end coordinates of your favourite gene (e.g. DNAH5)
How many exons does it have?
How many different transcripts does it have?
What is the function?
Associated with any disorders?
Which proteins are predicted to interact with it?
Extract the coding sequence – in FASTA

29. Answers
Chr 5 from 13,690,440 to 13,944,652 79 4
Force generating protein of respiratory cilia (from GeneCards – UniprotKB section)
Primary ciliary dyskinesia
DNAH1, DYNLL1, DNAL4 (from STRING)
Many ways to do, for example:
Export data in Ensembl (DNA)
Q8TE73 in Uniprot (AA)

30. Thank You Any questions?
Please look back at the slides again once you complete the short-course(s)

31. Appendices
Two additional terms which must be understood to make full use of the databases
10 useful websites/databases we do not have the time to go through
Two additional ‘must know’ data formats

32. LD Non random association of alleles at two or more loci
Simple example:
A at chr1:1000 and T at chr3:500
C at chr1:1000 and A at chr3:500
No other haplotypes
Therefore chr1:1000 and chr3:500 are in LD
Thus all SNPs are not independent
Therefore carefully selected SNPs can save money (e.g. Tag SNPs)
Different from linkage
Proximal loci on the same chromosome inherited together during meiosis
These designed SNP arrays were also used in the HapMap project to discover LD in different populations
If you know that 50 SNPs are inherited together (as a block), then you can just genotype one of them and impute the rest – more savings

33. Genetic association study
GWAS
SNP arrays - using LD
Very cheap (23andme offers for $100)
Case v Controls
Whole exome sequencing (WES)
Whole genome sequencing (WGS)
Expensive
Candidate gene analyses
E.g. ones identified by GWAS
Animal models
Where thousands of individuals are needed to detect the effect of a SNP on a phenotype, GWAS are used. If very few people are enough then WGS or WES is used.
Anything that is in LD with the SNP can be the causal one. That SNP you’ve genotyped just represents the whole LD block

34. RefSeq (NCBI) http://www.ncbi.nlm.nih.gov/refseq/
“A comprehensive, integrated, non- redundant, well-annotated set of reference sequences including genomic, transcript, and protein”
>33000 species (and/or strains)
CCDS project

35. GWAS catalogue www.genome.gov/gwastudies/
Search for SNP-phenotype associations from GWAS
Search, view and filter
Try example:
BMI and P value of 1e-8
Result: 11 papers (as of 21/02/14)

36. HGMD http://www.hgmd.cf.ac.uk/ac/index.php
“HGMD represents an attempt to collate known (published) gene lesions responsible for human inherited disease”
Public version
registration required
Professional version
Purchase a licence

37. PDB – Protein Data Bank http://www.rcsb.org/pdb/home/home.do
Links biochemistry to your study
If there is data of course!
3D view of protein (in Jmol)
Amino acid sequence
Try example:
FTO (4IDZ)

39. Pfam http://pfam.sanger.ac.uk/ Protein families and domains
Predicted to have similar functions
Domain organisation
Phylogenetic tree
Links to PDB
Try example:
Dynein_heavy (PF03028)

40. STRING http://string-db.org/ Predicts protein-protein interactions
Coexpression
Literature
Experiments
Genomic context
Try example:
DNALI1

42. KEGG http://www.kegg.jp/ Similar to STRING but manually curated
“resource for understanding high-level functions and utilities of the biological system, such as the cell, the organism and the ecosystem, from molecular-level information, especially large-scale molecular datasets generated by genome sequencing and other high- throughput experimental technologies”
Similar to STRING but manually curated
More reliable
My friends tell me the difference between STRING and KEGG is like the difference between SWISS-Prot and TrEMBL
STRING is automated, KEGG is manually curated – more reliable but less information

43. Regulatory elements Rfam (RNA family) Noncoding RNA database
Noncoding RNA database
Bioexplorer.net

44. ENCODE http://www.nature.com/encode/
Radical shake up to the interrogation of genomic “function”
Data available
Functional impact of variant sites in multiple tissues
Multiple assay types.
Analysis/visualisation software and scripts for the generation of figures
Massive database – won’t be going through it
No time for it either
However the good news is that Ensembl has direct links to it and is annotates some variations in accordance with ENCODE (regulatory feature)

46. Locus specific or disease specific databases
By HGVS
Human Genome Variation Society
Example:
Ciliome database

47. PED/MAP Most used format User friendly software Ped: Map: Plink
pngu.mgh.harvard.edu/~purcell/plink/
One of the mostly cited software
Ped:
FAM A A G G A C ...
FAM A A A G
....
Map:
1 rs
1 rs
1 rs …

48. Variant effect predictor (VEP)
Example file: