L6: Characterizing genetic diversity

how to identify genes of interest?

causal genes are gene that is responsible for variation in a phenotypic trait or a disease of interest

at meiosis, crossing overs swap - between - chromosomes, without recombination, all positions in a chromosome would be in -

recombination breaks the linkage between - in a chromosome

recombination mixes up - and creates new - combinations

regions. homologous, linkage

genomic positions

genetic variants, allelic

crossing over generates new combinations of the -

regions common to all progeny with the - phenotype will contain the -

parental chromosomes

mutant, causal gene

qualitative traits

• discrete classes

• can be categorical

quantitative traits

• continuous variation

• no discrete classes

• measurable

qualitative traits are typically -

meaning what?

monogenic

gene at one locus controls trait

quantitative traits are typically -

meaning what

polygenic

combined effects of many loci

to map the phenotypic effects of a qualitative trait, track segregation of only -

one gene/locus

when mapping quantitative variation, track segregation of -

QTL mapping

several genes/loci

QTL mapping

• each genetic region/variant is tested for associations with the phenotypic trait of interest

• the likelihood of that marker/variant being associated with a gene controlling that trait is plotted across the genome

what is needed for QTL mapping?

segregating population

• a group of related individuals obtained by crossing two or more parents, which shows variation for a trait of interest

genotype information

• genetic markers differentiating the parental genotypes across the whole genome

phenotype information

• measurements of trait of interest, as accurate and quantitative as possible

genetic mapping relies on

recombination breaking the association between alleles in nearby regions of the genome

the more recombination events you have, the higher the likelihood to break linkage between nearby genes

can narrow the candidate region further, meaning higher mapping resolution

therefore fewer possible candidate genes to choose from

how many recombination events?

on average, only 1-2 recombination per chromosome per generation

limited number of recombinations in each individual

recombination rates increase with chromosome number, not genome size

F2 characteristics

• 2 parents

• “quick” to produce

• limited recombination

• heterozygous regions

• one-and-done

RIL

• 2 parents

• require multiple generations

• more recombination

• homozygous individuals

• immortalized lines

MAGIC populations

multi-parent advanced generation inter-cross

• many parents = more alleles

• more recombination

• homozygous, immortalized lines

• very slow/work-intensive to produce

reference genome allows

placing and comparing of sequence reads from different individualshow

how to compare genetic diversity with short reads?

for genotyping, throughput and cost of sequencing is more important than read length

2nd gen seq produces large amounts of short reads for cheap

• short reads are great to identify genetic variants between individuals

• ideally comparing them to a reference assembly

limited to

• small variants like SNPs or indel

• non-repetitive region

how is read QC done?

1. trimming

2. mapping to reference genome

3. removing duplicates, indel realignment, base recalibration

4. variant detection

5. variant filtering

trimming

to remove low-quality bases and/or adapters

if inserts are smaller than 150bp, adapters will be sequenced as well

mapping to reference genome + removing duplicates

reads from each individuals are associated to the corresponding region on the reference assembly

reads from repetitive sequences will map to multiple regions of the genome, which are then discarded

variant calling/detection

positions that are variable (polymorphisms) are identified and collected in a variant call format (VCF)

variant filtering

assessing and improving the quality of raw sequencing reads, using that information to filter out or refine low-confidence variant calls generated from those reads

whole genome sequencing WGS

• provides very dense, genome-wide information on genetic differences between individuals

• cheap

• stumped by repetitive regions or variants larger than a few base pairs

WGS is not same as genome assembly

WGS = producing sequencing reads from the whole genome

genome assembly = reconstructing the sequence and organization of the whole genome

don’t need to know every single variant to assemble a genome

• genome-wide patterns of differentiation

• high levels of linkage

• limited number of crossing overs in an F2 population, so we don’t need to know the genotype of the individuals at every single nucleotide

recombination vs marker density for genetic mapping

recombination is a limiting factor in determining QTL mapping accuracy

even perfect knowledge of genotypes would not pinpoint the causal gene

not sequencing the whole genome allows

• save time and computational resources

• save money

• focus on regions that could be more interesting to you

reduced representation in regards to gene space

RNAseq

• potentially all genes, no previous knowledge required

• but affected by expression levels and RNA is delicate

hybridization seq

• selected regions, gene space and less seq needed

• but seq knowledge required and its an initial investment

reduced-representation in regards to genotype by sequencing

WGS sequences the whole genome

GBS sequences only a subset of genomic regions

• restriction enzymes recognize and cleave short DNA patterns

short reads genotyping focuses on -

SNPs

small but mighty

• missense/nonsense

• affects splicing

• disrupt/modify promoter sequence

• disrupt/add miRNA target sites

how to study structural variants?

short reads are poor at detecting structural variants

long reads allows identification of larger structural variants

getting phenotypic information from genotypic information

environment needs to be controlled to make sure only genetic factors are identified

genotypic information is increasingly easy/cheap thanks to new sequencing technologies

phenotypic information needs to be as accurate as possible

molecular phenotypes

phenotypes don’t need to be visible

• physiological states like temperature, transpiration, water content

• chemical or metabolite levels

• expression levels eQTL

• methylation levels mQTL

repeating QTL experiments in different conditions can identify genes controlling responses to -

environmental stimuli

what determines the success of a mapping experiment?

power = ability to detect genotype-phenotype associations

more power = genes with smaller effects can be detected

• size of population, more robust assct, more recomb

• complexity of trait, highly polygenic traits require more power

• heritability of trait, how much variation is due to genetics vs environment

• phenotype accuracy, sloppy phenotyping will mess up everything

• genotyping density, enough genotype information