Sequencing II: Bioinformatics

NM_004006.2

Reference sequence

-Where the variant is in relation to the reference sequence used

c.

Letter prefix

124A>G

Position and type of change

What do reference sequences start with?

NG

Genome builds

36.1/hg18, GRCh37/hg19, GRCh38/hg38

What do reference mRNA sequences start with?

NM

What do reference protein sequences start with?

NP

Why is the correct reference sequence important?

-Genes can have multiple isoforms

--Different isoforms of a gene will have different reference sequences

-Alternative splicing; gives us a different transcript that results in a different protein

Genomic DNA (g.)

First nucleotide of the genomic reference sequence

Coding DNA (c.)

First nucleotide of the translation start coding on the coding DNA reference sequence

Noncoding DNA (n.)

First nucleotide of the noncoding DNA reference sequence

Mitochondrial DNA (m.)

First nucleotide of the mito DNA reference sequence

RNA (r.)

First nucleotide of the translation start codon of the RNA reference sequence, or first nucleotide of the noncoding RNA ref seq

Protein (p.)

First amnio acid of the protein sequence

CNV- Deletion (del) format

"prefix""position(s)_deleted""del"

CNV- Duplication (dup) format

Only used for tandem duplication

"prefix""position(s)_duplicated""dup"

CNV- Insertion (ins) format

"prefix""positions_flanking""ins""inserted_sequence"

CNV- Inversion (inv) format

"prefix""positions_inverted""inv"

Contig

Historical term when genome was first sequenced

"Chunks" of sequence that is your 'reference' point relative to the g.___ position

Duplicate reads

Generated during PCR amp in library prep

-Too many can skew variant calling algorithms and introduce PCR-based sequencing errors into the variant calling algorithms

Depth

The number of sequencing reads at a given locus

-How many unique seq reads at a given locus?

Breadth

The amount of target loci with sequences aligned

-How much of the genome are we covering?

Possible issues w/ calling SNVs

-Zygosity

-Mosaicism

-Seq errors that are called as variants

Variant calling algorithms

Predict the likelihood of a variant vs. the likelihood of an artifact

-Quality score of individual reads

-Allele counts at the locus

-Strand bias

-Repeated regions

-Low complexity regions

More high quality reads w/ the same allele, the greater the likelihood that the variant is called

Coverage depth

Important to ensure accuracy of variant calling, especially in sample w/ low allelic fractions

-The more reads you have, the more likely it's a true variant!

What are some circumstances where you may detect a

low fraction of reads with an alternative allele?

1. Low level mosaicism

2. Tumor heterogeneity

Allelic fraction cut-off for mosaicism?

20% or lower

Allelic fraction of G:100%

Genotype is G/G

Homozygous for alternative allele

Allelic fraction of G:60% and A:40%

A/G

Heterozygous for alternative allele

Allelic fraction of G:20% and A:80%

A/G

Heterozygous BUT w/ mosaicism

Deeper coverage = ?

Higher quality variant calling

Calling structural variants

Measure relative read depth to infer changes in copy number

Less than average depth = ?

Deletion

More than average depth = ?

Duplication

Split reads

A single read where the 5' and the 3' end alignment do not align contiguously in the genome

Mapping paired reads- Above average distance between paired end reads = ?

Deletion

Mapping paired reads- Below average distance between paired end reads = ?

Insertion

FASTQ

Raw sequence reads and quality scores directly from the sequencer

BAM

Position information of sequence reads aligned to the reference genome, along w/ quality info

VCF

Variant call format

-Every location the sample differs from the reference

Filters to identify disease-causing variants exclude

1. Intronic or intergenic variants (not within exon)

2. Common variants

3. Synonymous variants

4. Variants not predicted to alter function

5. Variants that don't fir inheritance pattern

Variant filtering

-Sorting through thousands of variants to find those that contribute to phenotype

-Filtering scheme depends on the sequencing method and phenotype/inheritance pattern

How would you filter to identify de novo variants?

Check parents

How would you filter to identify variants causing an AR condition?

-Are parents affected?

-Do they have 1 or 2 copies?

-Does child have 2 bad copies?

How would you filter to identify variants causing an X-linked recessive condition in a male?

See if mom is a carrier of the condition