This document provides an overview of variant analysis from next-generation sequencing data. It begins with introductions to the CCA-Drylab@VUmc, TraIT, and Galaxy projects. The focus of the lecture is explained to be variant analysis from NGS data using interactive demos in Galaxy. Background is provided on Illumina sequencing technology and properties of sequencing reads. Key steps in variant analysis are outlined, including quality control and read mapping, variant calling and annotation using tools like FastQC, BWA, FreeBayes, and SnpEff. Formats for storing sequencing data and variants are also introduced, such as FASTQ, SAM/BAM, and VCF.

1. Introduction to Variant Analysis
of Exome- and Amplicon sequencing data
Lecture by:
Date:
Dr. Christian Rausch
26 April 2016

2. About CCA-Drylab@VUmc, TraIT, Galaxy Project
• CCA-Drylab at the VU University Medical Center Amsterdam is the
Bioinformatics initiative of the Cancer Center Amsterdam (CCA). One goal
is to give trainings in bioinformatics and make workflows available to
researchers. Galaxy was selected as a tool to achieve this.
• TraIT (Translational Research IT) is a Dutch national project which is part of
CTMM, the Center for Translational Molecular Medicine.
• TraIT is a scientific IT project with the goal to enable integration and
querying of information across the four major domains of translational
research: clinical, imaging, biobanking and experimental (any-omics) with
a particular focus on the needs of multi-center projects.
• Galaxy is one of the tools supported and extended by TraIT
• Galaxy is an international project lead by teams at Pen State and Johns
Hopkins University (15 FTEs working on the development). It is a graphical
bioinformatics analysis platform aimed at Computational Biologists.

3. Focus of lecture and
practical part
• Lecture: from NGS data to Variant analysis
• Live demo: we will analyze NGS read data of
the reference NA12878 from “Genome in a
Bottle”/GCAT. DNA Analysis software tools
will be run interactively through Galaxy, “a
web-based platform for data intensive
biomedical research”
Image source: A survey of tools for variant analysis of next-generation
genome sequencing data, Pabinger et al., Brief Bioinform (2013) doi:
10.1093/bib/bbs086

4. Recap on Illumina’s sequencing technology

5. Recap sequencing (Illumina technology) slide 1
Image source: openwetware.org/images/7/7a/DOE_JGI_Illumina_HiSeq_handout.pdf

6. Recap sequencing (Illumina technology) slide 2
Image source: openwetware.org/images/7/7a/DOE_JGI_Illumina_HiSeq_handout.pdf

7. Recap sequencing (Illumina technology) slide 3
Image source: openwetware.org/images/7/7a/DOE_JGI_Illumina_HiSeq_handout.pdf

8. Recap sequencing (Illumina technology) slide 4
To be able to sequence paired reads, a second
bridge amplification followed by a ‘flip’ of the
template is required
Image source: illumina.com

9. Properties of
Reads (Illumina)
• Typical read length: 50 … 100 … 150 …
200 … 300 bp
• Read quality drops after 100, 150 or
200 bp (depending on model of
sequencer)
• Paired reads: Insert size 200 – 500 bp
• Mate pairs: Insert size several kbp
• Errors in Illumina reads are typically
substitution errors

10. Properties of
Reads (Illumina)
• Paired reads: Insert size 200 – 500 bp
• Mate pairs: Insert size several kbp Source:
evomics.org/2014/01/alignment-
methods/
Image
source: Mate
Pair v2
Sample Prep
Guide For 2-5
kb Libraries

11. Comparison Sequencing Platforms
Platform Amplification
Method
Sequencing Method Detection Method Average read length
Illumina (HiSeq,
MiSeq etc)
bridge PCR sequencing by
synthesis
Light 100-200 bp
Life Tech Ion Torrent
/ Proton
emulsion PCR Ion semiconductor
sequencing
pH 200-400 bp
Roche 454 emulsion PCR Pyrosequencing,
cleavage of released
pyrophosphate
light 700 bp
Life Tech SOLiD emulsion PCR sequencing by
ligation of
hybridizing labeled
oligos
light 100 bp
Pacific Biosciences
PacBio / Sequel
System
No amplification,
single-molecule
sequencing
polymerase
incorporating
colored NTPs
light 10 kb, 50% >20 kb,
5% > 30 kb
Oxford Nanopore
MinION
No amplification,
single molecule
nanopore
sequencing
DNA molecule
traverses pore
current > 5.4 kb
Further reading, great lecture: Sequencing technology - Past, Present and Future, http://www.molgen.mpg.de/899148/OWS2013_NGS.pdf

12. MinION: Oxford Nanopore’s “3rd generation” Sequencer:
Image source: John MacNeill,
http://www2.technologyreview.com/article/427677/nanopore-
sequencing
• Based on biological nano-machines,
reading single DNA molecules
• DNA Sequencer ‘MinION’ size of USB
drive
• Extremely fast (first results ~1h)
• Much longer reads than currently
established sequencing technologies
 Puzzle with large parts is easier
than with small ones
• Used by more and more leading labs
Example of a relevant recent publication:
http://publications.nanoporetech.com/2015/10/05/nanopore-sequencing-detects-structural-variants-in-cancer/

13. Target enrichment
used for selecting
exomes
• Image source: Agilent

14. Selecting parts of the genome for sequencing
• The Illumina TruSeq Amplicon -
Cancer Panel uses Multiplex
PCR to amplify a selected part
of the genome (a selection of
the exons of 48 genes are
targeted with 212 amplicons)

15. Intro Bioinformatics: NGS based Variant Analysis

16. Workflow: QC & Mapping reads
Input reads
(fastq files)
Quality check
with FastQC
Quality- & Adapter-
trimming
Not OK?
OK?
Map reads to reference
genome
using e.g. BWA or Bowtie2
Output:
Sorted BAM file (binary SAM
sequence alignment map)
Galaxy sorts mapped reads by
coordinates using samtools sort
Mark reads that are likely
PCR duplicates (Picard tools
Mark Duplicates)
Source: clc-bio.com
PCR Deduplication

17. Variant Calling & Annotation pipeline
Reads
mapped to
reference
genome
SAM or BAM
file
Freebayes
• Analyze mismatches & compute likelihoods of SNP etc.
• Call variants(decide if position is altered or not, based on statistical model)
• Can be limited to region of interest
• Output: VCF
• Various statistics on quality of each variant (read depth
etc.), homozygous/heterozygous etc.
Filter variant
allele frequency
Discard variants
with variant
allele frequency
below threshold
SnpSift Annotate:
Annotate known
Variants with
dbSNP/Clinvar db
etc.
SnpEff: Effect annotation
Genetic variant
annotation and effect
prediction toolbox. It
annotates and predicts
the effects of variants on
genes (such as amino acid
changes).

18. Quality Measure: Phred Score
Phred quality score
Probability that the base
is called wrong
Accuracy of the base call
10 1 in 10 90%
20 1 in 100 99%
30 1 in 1,000 99.9%
40 1 in 10,000 99.99%
50 1 in 100,000 99.999%

19. Quality Measure: Phred Score
• Phred score = quality scores
• originally developed by the base calling program Phred used with Sanger
sequencing data
• Phred quality score Q is defined as a property which is logarithmically
related to the base-calling error probability P
• Error rate P = 10 – (Phred score Q / 10)
• Q = -10 log10 P
• Example: Phred score 30 = error rate 10-3 = 1 base in 1000 will be wrong
• Illumina’s ‘Q score’ = Phred score = Quality score in Sanger sequencing
• The base calling programs that convert raw data to sequence data (the
‘base callers’) need to be ‘trained’ to give realistic quality values

20. • format
standard format to store sequence data (DNA and protein seq.)
>FASTA header, often contains unique identifiers and descriptions of the sequence
@unique identifiers and optional descriptions of the sequence
the actual DNA sequence
+ separator optionally followed by description
The quality values of the sequence
(one character per nucleotide)
standard format to sequencing reads with quality information
(‘Q’ stands for Quality)
More info see wikipedia ‘FASTQ_format’

21. Quality control
with FastQC
• Need to check the
quality of reads
before further
analysis
• Program FastQC is
quasi standard
• Sequencing
platform
companies provide
also their own
tools for quality
control

22. Quality control: FastQC
• Encoding tells which set of
characters stands for which
Phred scores.
• There’s also Encoding Illumina
1.5 and others.
• Other programs might not
automatically recognize the
encoding
• In Galaxy there is a possibility to
set the encoding of a FastQ file
via the ‘pen’ symbol.

23. Read Quality Control with FastQC
Examples of per base sequence quality
of all reads
 Historical example of very first Solexa
reads 2006 (Solexa acquired by Illumina 2007)
Not so good, might still be usable,
depending on application
50 bp

24. Examples of
other
quality
measures in
FastQC
• Upper 4 graphs
from the data set
of the practical
course:
• Many reads are
repeated
• Apparently not
uniformly
distributed over
whole genome
• Overrepresented
sequences:
Sequenced
fragment was too
short and
sequencing
reaction ran into
the Adapter/PCR
primer

25. “Mapping reads to the reference” is
finding where their sequence occurs in the genome
Source: Wikimedia, file:Mapping Reads.png
100 bp identified 100 bp identified200 – 500 bp unknown sequence

26. “Mapping reads to the reference”:
naïve text search algorithms are too slow
• Naïve approach: compare each read with every position in the genome
– Takes too long, will not find sequences with mismatches
• Search programs typically create an index of the reference sequence (or
text) and store the reference sequence (text) in an advanced data
structure for fast searching.
• An index is basically like a phone book (with
people’s addresses)  Quickly find address
(location) of a person
Example of algorithm using ‘indexed
seed tables’ to quickly find locations
of exact parts of a read

27. “Mapping reads to the reference”: frequently used
programs
• BLAST, the most famous bioinformatics program since 1990, is used to find
similar sequences in DNA and protein data bases
– 0.1-1 sec to find a result
– Mapping 60 million reads would take ~ 2 months on one CPU1  too slow
for NGS
• Popular tools for read mapping:
– Bowtie, Bowtie2, BWA, SOAP2, MAQ, RMAP, GSNAP, Novoalign, and
mrsFAST/mrFAST: Hatem et al. BMC Bioinformatics 2013, 14:184
http://www.biomedcentral.com/1471-2105/14/184
– CLCbio read mapper (commercial)
– No tool is the best tool in all example conditions
– differences in speed
– Differently optimized for mismatches/gap models/Insertions &
Deletions/taking into account read base quality/local realignment of matches
etc.

28. Read Mappers: BWA and Bowtie2
Are based on the Burrows-Wheeler Transformation (BWT)
• BWT: special sorting of all letters in the text (sequence)
• Similar suffixes (word ends) will be close to each other
• Easier to compress
• Good for approximate string matching (sequence alignment)
• Index (FM index) for finding the locations of matched strings (sequences)
in the genome

29. Read Mapping: General problems
• Read can match equally well at more than one location (e.g. repeats,
pseudo-genes)
• Even fit less well to it’s actual position, e.g. if it carries a break point,
insertions and/or deletions

30. SAM and BAM files
• SAM = Sequence Alignment Map
• BAM = Binary SAM = compressed SAM
• Sequence Alignment/Map format
• contains information about how sequence reads map to a reference
genome
• Requires ~1 byte per input base to store sequences, qualities and meta
information.
• Supports paired-end reads and color space from SOLiD.
• Is produced by bowtie, BWA and other mapping tools
Partly from: genetics.stanford.edu/gene211/lectures/Lecture3_Resequencing_Functional_Genomics-2014.pdf

31. SAM Format
Example from: genetics.stanford.edu/gene211/lectures/Lecture3_Resequencing_Functional_Genomics-2014.pdf
Example Alignment:
Encoding SAM format:

32. Harvesting Information from SAM
• Query name, QNAME (SAM) / read_name (BAM).
• FLAG provides the following information:
– are there multiple fragments?
– are all fragments properly aligned?
– is this fragment unmapped?
– is the next fragment unmapped?
– is this query the reverse strand?
– is the next fragment the reverse strand?
– is this the last fragment?
– is this a secondary alignment?
– did this read fail quality controls?
– is this read a PCR or optical duplicate
Source: www.cs.colostate.edu/~cs680/Slides/lecture3.pdf

33. Variant Calling & Annotation

34. Possible reasons for a mismatch
• True SNP or Mutation
• Error generated in library preparation
• Base calling error
– May be reduced by better base calling methods, but cannot be eliminated
• Misalignment (mapping error):
– Local re-alignment to improve mapping
• Error in reference genome sequence
Partly from www.biostat.jhsph.edu/~khansen/LecSNP2.pdf

35. Variant Calling: Principles
• Naïve approach (used in early NGS studies):
– Filter base calls according to quality
– Filter by frequency
– Typically, a quality Filter of PHRED Q 20 was used (i.e.,
probability of error 1% ).
– Then, the following frequency thresholds were used according to
the frequency of the non-ref base, f(b):
– The frequency heuristic works well if the sequencing depth is
high, so that the probability of a heterozygous nucleotide falling
outside of the 20% - 80% region is low.
– Problems with frequency heuristic:
• For low sequencing depth, leads to undercalling of heterozygous
genotypes
• Use of quality threshold leads to loss of information on
individual read/base qualities
• Does not provide a measure of confidence in the call
In parts from: compbio.charite.de/contao/index.php/genomics.html

36. Variant Calling: Principles
• Today’s Variant Callers rely on probability calculations
• Use of Bayes’ Theorem:
– E.g. MAQ: One of the first widely used read mappers and
variant callers
– See introduction at:
https://en.wikipedia.org/wiki/SNV_calling_from_NGS_data
• Takes into account a quality score for whole read
alignment & quality of base at the individual position
• Calls the most likely genotype given observed substitutions
• Reliability score can be calculated

37. Variant Calling & Annotation: Popular Tools
• SAMtools (Mpileup & Bcftools)
• GATK
• Freebayes
• Varscan2
• Mutec
• MAQ (historic)

38. VCF = Variant Call Format
• Variant Call Format / BCF = binary version

39. Variant Annotation with: dbSNP and snpEff
• dbSNP = the Single Nucleotide Polymorphism Database @ NCBI
• Different collections of SNPs are available: ‘all humans’, human
subpopulations, different clinical significance
(www.ncbi.nlm.nih.gov/variation/docs/human_variation_vcf).
• snpEff is a program that can annotate a collection of SNVs according to
information available in dbSNP and information extracted from the location
of the SNV (Exon, Intron, silent/non-sense mutation etc.)

40. Variant annotation with: ANNOVAR
a versatile tool to annotate genetic variants
(SNPs and CNVs)
• Input:
– variants as VCF file
– various databases with statistical and predictive annotations: dbSNP, 1000 genomes, exome variant
server (Univ. of Washington), …
• Output:
– In coding region? Which gene? How frequently observed in 1000 genomes project? (and more
statistics).
• According to coordinates from RefSeq genes, UCSC genes, ENSEMBL genes, GENCODE genes or
others, etc.
– In non-coding region? Conserved region?
• According to conservation in 44 species, transcription factor binding sites, GWAS hits, etc.
– Predicted Effect?
• SIFT predicts whether an amino acid substitution affects protein function. SIFT prediction is
based on the degree of conservation of amino acid residues in sequence alignments derived
from closely related sequences, collected through PSI-BLAST. SIFT can be applied to naturally
occurring nonsynonymous polymorphisms or laboratory-induced missense mutations
• PolyPhen-2 (Polymorphism Phenotyping v2) is a tool which predicts possible impact of an
amino acid substitution on the structure and function of a human protein using straightforward
physical and comparative considerations.
• PhyloP: Detection of nonneutral substitution rates on mammalian phylogenies.

41. Annotation with DGIdb: mining the druggable genome
Drug Gene Interaction Database:
Matching disease genes with potential drugs.
• Searches genes against a compendium of drug-gene interactions and identify
potentially 'druggable' genes
• DGIdb 2.0 Overview:
• New Sources of Drug-Gene Interactions
– CIViC, an open access, open source, community-driven web resource for Clinical Interpretation
of Variants in Cancer.
– DoCM, the Database of Curated Mutations, a highly curated database of known, disease-
causing mutations.
• Automatic updating of sources
– CIViC - drug-gene interactions
– DoCM - drug-gene interactions
– DrugBank - drug-gene interactions
– Gene Ontology - druggable gene categories
– Entrez Gene - genes
– PubChem - drugs
• Search drug-gene interactions by drug identifiers
– Go to the search interactions page and click on the Drugs button to try it out

42. Genome in a Bottle Consortium & GCAT
Genome in a Bottle Consortium hosted by the National
Institute of Standards and Technology, (NIST), is creating
reference materials and data for human genome
sequencing, as well as methods for genome comparison
and benchmarking
Example:
• Sequenced ‘pilot genome’ NA12878 with 11 sequencing
technologies.
• Variant reference set created with seven read mappers
and three variant callers
Genome Comparison and Analytic Testing
(GCAT) platform wants to facilitate
development of performance metrics and
comparisons of analysis tools across these
metrics. Provides interactive visualizations of
benchmark and performance testing results.

44. Practical part (Galaxy live demo)
Note: Public Galaxy Servers and how to install Galaxy yourself:
• https://galaxyproject.org/