Identification of Copy Number Variants using Genome Graphs

1
Identification of Copy Number Variants using
Genome Graphs
Dhawal Verma Advisor Dr. Hesham Ali
2
Introduction

• The genome of an organism offers great insight
  into its
• phylogenetic history
• interaction with the environment
• internal functions
• Even within the same species, the genomes of two
  individuals differ. Although the genomic
  variations are relatively small, they account for
  the observed variations in
• Phenotypes (Heterozygosity)
• Susceptibility towards various diseases.

3
Motivation

• Heterozygosity is of major interest to
  researchers of genetic variation in natural
  populations.
• It refers to the state of having different
  alleles at one or more corresponding chromosomal
  loci.
• It is often one of the first "parameters" that
  one presents in a data set. It can tell us a
  great deal about the structure and even history
  of a population.

4
Motivation

• Role in diseases
• SVs and CNVs have been associated with
  susceptibility or resistance to disease.
• Gene copy number can be elevated in cancer cells.
• Copy number variation has also been associated
  with autism, schizophrenia and idiopathic
  learning disability.

5
Visualization of Genome

• Genome A Book
• Written in 4 letters of nucleotides A T G C
• 23 Chromosomes 23 Chapters
• Genes Stories in each chapter

6
Genome
A T G C
7
Genomic Structural Variation

• Every Genome differs from another, however like
  different books differ from one another, the list
  of words used in the book comes from a known
  dictionary of words.
• Like different positions of various words in a
  sentence give out a different meaning, different
  positions of the same gene in a genome give us a
  distinct feature and causes a variation in
  genomes.

8
Genomic Structural Variation

• Until fairly recently, single nucleotide
  polymorphisms (SNPs) were thought to be the main
  source of variation in the human genome.
• SNPs are variations that involve a change in just
  one nucleotide.
• THE RAT CAN RUN FAST
• THE CAT CAN RUN FAST
• High-throughput genome scanning technologies
  revealed that there are other forms of genomic
  variation beyond single base-pair substitutions.

9
Structural Variants

• Structural variant is the umbrella term to
  encompass a group of genomic alterations
  involving segments of DNA typically larger than 1
  kb.
• The structural variation may be
• Quantitative (CNVs indels and duplications)
• Positional (translocations)
• Orientational (inversions).

10
Copy Number Variants (CNVs)

• CNVs are defined as chromosomal segments, at
  least 1000 bases (1 kb) in length that vary in
  number of copies from human to human.
• CNVs are large chunks of DNA that are deleted,
  copied, flipped or otherwise rearranged in
  combinations that can be unique for each
  individual.
• YOU CAN RUN FAST
• YOU CAN RUN RUN RUN FAST

11
SNP v CNV

• SNPs always occur in two alleles, while
  approximately 5 of the human genome are defined
  as structurally variant in the normal population,
  involving more than 800 independent genes.
• Of the total amount of variation between two
  human individuals
• CNVs SVs gtgtgt SNPs

12
Primitive methods for detection of CNVs

1. Whole-genome array comparative genome
   hybridization(aCGH), which tests the relative
   frequencies of probe DNA segments between two
   genomes
2. SNP arrays to measure the intensity of probe
   signals at known SNP loci.

13
Limitations of the methods

• The size and breakpoint resolution of any
  prediction is correlated with the density of the
  probes on the array, which is limited by
• the density of the array itself (for aCGH)
• the density of known SNP loci (for SNP arrays).
• The limited resolution of arrays for high copy
  count segments and the lack of unique probes make
  it difficult to identify CNVs in repetitive
  regions.

14
Research Proposal

• An effective computational method for the
  identification of Copy Number Variants in
  genomes.
• Model
• Next generation sequencing data can be modeled in
  a graph that we call a Genome Graph
• Algorithm
• By effectively mapping the reference genome graph
  with the donor graph and making use of two
  different existing methods known as Depth of
  coverage and Paired end mapping together, we can
  overcome their limitations and detect the CNVs
  with higher sensitivity and specificity.

15
Research Proposal

• Our literature survey indicates that PEM method
  is used specifically for detecting SVs and DOC
  method for CNVs.
• CNVs in general are considered as a subset of
  SVs.
• By integrating the two methods we can use PEM
  signatures at a higher magnification level.
• Also the complexity can be reduced by using the
  bi-directional genome graphs.

16
Genome Graphs

• With the advent of Next Generation Sequencing
  data that provides as much as 40x coverage for a
  human genome, a special class of graphs known as
  Genome graphs emerged.
• The vertices represent either the reads or their
  substrings (k-mers expressed by various
  combinations of the letters A,T,G and C)
• The edges represent overlaps between them (the
  prefix of one read is the suffix of the other).

17
Genome Graphs

• A genome graph can be unidirectional or
  bi-directional.
• Bi-directional genome graph implements the
  double-strandedness of DNA.
• Bi-directional graphs help reduce the complexity
  of algorithm as in unidirectional graphs two
  complementary walks are searched while in
  bi-directional graph a single walk can fetch both
  the sequence and its complement.

18
Depth of Coverage method

• Depth of Coverage
• The density of reads mapping to the region
• Several recent studies have shown that by
  comparing the DOC within a sliding window of the
  genome to what is expected in the reference
  genome, it is possible to detect changes in copy
  number
• Limitations
• Very Complicated
• difficult to separate true changes in copy number
  from segments that are over or under sampled by
  the sequencing technology.

19
Depth of Coverage
In a genome graph, an increase/decrease in number
of vertices between two known vertices in the
reference genome gives an indication of CNV.
20
Paired End Mapping method

• PEM method
• two paired reads (called matepairs) are generated
  at an approximately known distance in the donor
  genome.
• The reads are mapped to a reference genome, and
  matepairs mapping at a distance significantly
  different from the expected length (termed
  discordant) suggest structural variants.
• Limitations
• Difficulty in detecting larger insertions and
  variation within areas of segmental duplications

21
PEM signatures in Genome Graphs
22
PEM signatures v DOC signatures

• In contrast to most PEM signatures, DOC
  signatures can be used to detect very large
  events.
• The larger the event, the stronger the signature.
• However, they are not able to accurately identify
  smaller events that PEM signatures, even with low
  coverage, are able to detect.

23
Next Steps

• While inversions do not cause any changes in copy
  number, an area that is deleted (SV) will
  correspond to a loss (CNV). Similarly, a region
  containing a tandem duplication will be annotated
  as both having an insertion (SV) and as
  exhibiting a gain (CNV). In this way, any PEM
  method for SV detection can be viewed as a method
  for detecting a subset of CNVs
• Depth of Coverage method is used extensively for
  detecting CNVs, PEM technique is majorly used for
  detecting SVs
• Our hypothesis is that PEM techniques can be used
  to improve both the sensitivity and specificity
  of depth of coverage based methods using a
  probabilistic graph-theoretic framework.

24
THANK YOU!