Genetic Testing:

Crash course in genomic

variant interpretation
Charles Billington MDPhD
Matthew Bower, MS-LGC
Genetics and Metabolism, Dept of Pediatrics
Molecular Diagnostic lab, Dept of Lab Medicine and
Pathology
No conflicts of interest or financial relationships to declare
Learning Objectives
Describe multiple types and levels of genetic variants and the tests commonly used to
identify genetic abnormalities. Includes NGS, Sanger
Identify some commonly used molecular techniques including CRISPR/Cas9, PCR,
Southern Blot, Northern Blot, Western Blot, Genotyping Microarray, Expression
microarray
Identify indications for genetic testing and rationales for select types of testing.
Identify indications for genetic testing and select tests commonly used to identify
genetic abnormalities.
Describe the bases of genetic test interpretation, including that genetic test
interpretation has specific kinds and strengths of evidence to assess
pathogenicity of genetic changes. Proces
s or
Recognize that genetic changes can be benign, disease causing (pathogenic),
of uncertain significance.

2
Who we are:
why us for this session?
Charles Billington Matt Bower
Medical Geneticist Genetic Counselor
Genetics researcher Molecular Lab Analyst
I see the patients and order the tests I also see patients directly and order
and make recommendations about genetic tests and guide medical team on
medical management. genetic testing and results
Collaborate on review of cases with In Molecular Diagnostic Lab we get
Molecular Diagnostic Lab samples from patient cases, send for
sequencing and analyze the data,
In research I also do sequencing and
interpreting variants found and issue
analyze genetic data.
reports.
Genetic Counselors initiate analysis
(later also reviewed by MD molecular
pathologist)
3
Why this presentation?
•Become better users of genetic information
for patient care
•Nearly all of you will encounter a genetic
result in your practice.
•Integrates information from multiple parts
of Fundamentals course

Proces
s

4
An analogy:

Everyone in medical school learns to be familiar with a chest X ray. we have experts in radiology to help with angiograms and
complex CT scans.

5
 Sanger NGS
• Low throughput (sequence • High throughput (sequence
one exon, one sample at a many exons, multiple samples
time) at a time) Millions of reads all
• Targeted at point of at once.
sequencing (primer) • Targeted by library prep or
• Qualitative (essentially) informatically
• Challenging to detect low • Quantitative (read depth)
allelic fraction variants • Enables non-targeted
• Somatic mosaicism, cancer etc. analyses, but more likely to
yield clinically uncertain
results

Note: neither method is optimal for repetitive regions

 Clinical Applications of NGS
• Targeted disease-specific gene panels
o Few to a couple hundred genes
o High coverage depth (hundreds)
• Whole exome (protein coding regions)
o ~20,000 genes
o Some exons may not be covered well
o Helpful when candidate genes or clinical diagnosis unclear
• Whole genome
o Includes exome (1-2% of genome) and everything else
o ~200 times more variants detected than for a whole exome test
o Just starting to be adopted clinically (cost, challenging to interpret)

o Not as deep of coverage

What test do you need?
• Known Family variant?
• Suspected syndrome with only one or a few genes for
which variants may cause the condition?
• Unknown condition strongly suggested to be genetic
based on clinical history or family inheritance?
• Patient has a clinical course outside the norm or
expected for that condition?
• Still needs a phenotype
What test do you need? Answers
• Known Family variant? Sanger
• Suspected syndrome with only one or a few genes for
which variants may cause the condition? NGS or
Sanger Panel
• Unknown condition strongly suggested to be genetic
based on clinical history or family inheritance? Exome
or Genome
• Patient has a clinical course outside the norm or
expected for that condition? Exome or Genome
Sample to Report (NGS):

https://www.genome.gov/genetics-glossary/Shotgun-Sequencing 10
Nomenclature for Variants:
small group discussion
Which term do we use when we have a change in a
patient? Why?
Variant
Mutation
Polymorphism

Verbally discuss with neighbor or with your small

group:
• What information would you need to have to
identify and communicate a specific change in a
DNA or gene sequence so that everyone would be
able to identify the same change?
11
Nomenclature for Variants:
Variant 👍, Mutation 👎, Polymorphism 👎 ,
Variant is best when we have a change in a patient
Mutation can be interpreted pejoratively
Polymorphism just means >1% in population

•Have to specify a location

and
•a change (include the reference & the change)
and
•indicate what the reference to interpret it is

Guidelines are at https://varnomen.hgvs.org/

12
 Most common
way to represent
Coding sequence coordinates variants in clinical
applications

• Specify gene, transcript, position, reference and change.

• Start with the DNA coding sequence processed mRNA but with T for U
• c. counts in coding sequence from the first A of the start ATG as “1”
• p. numbers the amino acids with first methionine as “1”
• Protein changes are usually in brackets since these are [predictions]
• c. often reported with the predicted protein change following in [ ]
• Can do with noncoding also with reference to nearest coding
base
• Intronic variants specified with closest coding base with + and –
• For coding or protein locations note which transcript is used
• For example: NM_001144967.3 (NEDD4L): c.623G>A [p.Arg208Gln]
and NM_001144966 (NEDD4L) c.260G>A [p.Arg87Gln] are the SAME
change

13
Genome Coordinates
• Specify chromosome, position, reference base, variant
base.
• For chromosomal locations, note which genome reference
assembly “Build”
• hg19 aka GRCh37
• vast majority of clinical testing is still reported with this reference
• hg38 aka GRCh38
• research articles mostly use this, it has significant improvements
• [hg19] chr18-55992337-G>A = [hg38] chr18-58325105G>A

**Note that the bases may be complementary to those in the

coding sequence description of the variant (genes can be in either
orientation on the chromosome)**
14
 So what does it all mean?
• Most variation is BENIGN • Some variation causes disease
• COLOR vs. COLOUR • CAT vs. BAT
• CENTER vs. CENTRE • PRESCRIBE vs. PROSCRIBE

• Some variation is of
uncertain significance
• You just can’t tell.
• You found a change, but
clinical impact unclear 15
Questions when examining variants:
● Does the patient’s phenotype fit with what is expected for this gene’s variants?
● Is this variant consistent with the expected mechanism of disease?
○ Eg Deficiency/Haploinsufficiency/Dominant Negative/Gain of Function
● Has this been reported in other databases? In literature reports?
● How common is the variant? Has it been reported in healthy individuals?
● What do in silico models predict?
● Is there genetic cosegregation data available?
● If recessive, is the variant seen in trans with another variant?
○ (Means that there is compound heterozygosity)
● Is there functional evidence?
○ In vitro characterization or clear clinical assay

16
Key paper for classification:
• Establishes widely used
standards and criteria
for classifying variants
• Establishes standardized
process for combining
these criteria to make an
overall assessment of a
variant
• Process still requires
clinical judgement so
different labs or analysts
might differ in
assessments.

Richards et al. 2015

PMID: 25741868
4 main types of evidence:
1. Effect on Protein
a) Variant type
b) Functional data
c) Location within protein
d) Computational
2. Population frequency
3. Inheritance/Genetics
data
• De novo?
• For Recessives: both
copies affected?
5. Other Reports:
a) Variant seen before
elsewhere?
b) Phenotypic match?

18
Translation: effect on protein?
• Synonymous TTG>CTG
• Missense GG GTG>ATG
• Null TAC>TAG
•Nonsense
Null variants are
•Splice usually strong
evidence of disease
•Frameshift causing potential

19
Some parts of a protein matter
more than others
• Repetitive areas can often • “Hotspots” in a protein
sustain loss of amino can be sites where almost
acids without impact any change will cause
disease
• Catalytic sites
• Sometimes revealed by
observed variation
• Places where another
clearly disease-causing
variant have been found
are more suspicious

Reference: ALD Mutation Database (https://adrenoleukodystrophy.info/) 20

Functional Evidence: In vitro or in vivo
• Can be based on invitro assays or
on clinical data that support the
putative mechanism
• Can be open to interpretation

Davids et al. PMID: 32165008 21

Alves et al. PMID: 30270084
Computerized
Predictions
• Many different computerized
predictors

• think about what computerized

predictions might be based on
and compare your reflection
with the next slide . . .

22
Computerized
Predictions
. . . are based on . . .

• Evolutionary conservation
• (DNA level, Protein level)
• Chemical Similarity of amino acid
side chains
• Proximity to known structural
elements
• Predicted structural/folding effects
• Machine learning on similar changes
• Aggregates of the above

23
Population frequency:
Genetic diseases are generally rare.

The alleles that cause them are usually not

common in the population

A pathogenic (disease causing) variant

should generally not be present at a high
frequency in healthy individuals

But: Benign variants can be rare too!

Anything with frequency > 5% in any

population is strong evidence of being
benign (stands alone). Anything with
frequency 1% or above is a polymorphism
and usually not pathogenic.
• note some rare exceptions to
frequency rules in certain populations
24
gnomAD:
most widely used population database
https://gnomad.broadinstitute.org

https://gnomad.broadinstitute.org/variant/7-117199644-ATCT-A?dataset=gnomad_r2_1

25
Who makes up these “populations”?
Note who is left out.
• Rarity means more to
interpretation the more sampled
a population is.
• Variants in undersampled
populations are more likely to
be reported as uncertain
significance

https://gnomad.broadinstitute.org/blog/2018-10-gnomad-v2-1/ 26
Does the gene track with the disease?
In families? In populations?
• Is it a NEW change? De novo inheritance in an affected
person is strong evidence for pathogenicity
• Does it track through a family with the disease?
• Is it seen in multiple families with the same condition?
• Recessives: is the change seen as compound heterozygote
with a known disease-causing allele.
• Is the change confirmed to be on the OTHER allele (“in trans”)?
• Is it seen seen significantly more often in affected
individuals?

27
Inheritance as evidence:
Co-segregation Inheritance in trans
Trans Cis

De Novo

28
What have others said?
• Published literature
• Sometimes contains enough information to make a full
independent assessment of variant
• ClinVar:
• Database contributed to by most clinical labs and any
variants published.
• Includes benign and deleterious variants.
• Variable in how reliable variant assertions are.
• Often not enough information to assess independently

https://www.ncbi.nlm.nih.gov/clinvar/

29
Putting it all together
• Online Tools for looking at
variants and adding up criteria
• Tools that aggregate variant
data from public databases and
provide automated calls based
on what is available from public
knowledge.
• https://www.medschool.umarylan
d.edu/Genetic_Variant_Interpretati
on_Tool1.html/
• https://varsome.com/
• https://franklin.genoox.com/

30
Richards et al. 2015
31
PMID: 25741868
 32
Nykamp et al. 2017, PMID: 28492532
Putting it all together - patient interpretation:
“The c.293C>T (p.Ser98Leu) variant in ABCD1 has been reported in multiple published patients with
X-linked adrenoleukodystrophy (PMID: 8651290, PMID: 11438993, PMID: 31074578, PMID:
22280810). In contrast, this variant is absent from the gnomAD control database. This variant has
been classified as pathogenic by clinical laboratories in the ClinVar database and in the X-linked
adrenoleukodystrophy mutation database. Published reports have provided evidence consistent
with a de novo origin for this mutation in some cases [PMID: 11438993] and have provided limited
evidence to suggest that the variant segregates with an X-linked ALD phenotype [PMID: 22280810].
Other variants involving this same codon (p.Ser98Pro and p.Ser98Trp) have been reported as
pathogenic in the X-linked ALD mutation database. Functional studies have demonstrated that his
variant does result in a stable protein, although this study could not assess whether the mutant
protein retained normal function [PMID: 8651290]. In silico models consistently predict a
deleterious effect for this variant, but the accuracy of these models is limited. Based upon the
available evidence, this variant is classified as pathogenic.”
[PM2] [PP5] [PM6] [PP1] [PM5] [PP3]

33
Exercise: YOUR TURN!
Variants from Molecular Diagnostic Lab at UMN or clinical practice
• What condition is the gene associated with?
• What does the change do to the protein?
• What is the allele frequency in gnomAD?
• Has anyone reported this change in ClinVar?
• If available, does family or inheritance information change your call?
• Is there functional evidence that informs your assessment?
• What do you think this variant’s call was in the lab?
• Does the patient have a diagnosis?

Start with Problem Set B in the handout from Canvas:

Problem sets A and C are helpful supplemental material.

34
Overall Summary:
small group discussion
• Most genetic variation is _________.
• Some gene variants cause disease
• There are specific criteria for assessing variants:
• Assessment of impact on __________?
• Frequency/______?
• Fit with the _________?
• _________ reports?
• Most of the time a VUS is truly uncertain
• Most VUSs will be eventually _________.
• Don’t say a VUS is an answer unless you are pretty sure.
• Sometimes a VUS really can be the answer though.
35
Overall Summary:
group share-compare
• Most genetic variation is benign.
• Some gene variants cause disease
• There are specific criteria for assessing variants:
• Assessment of impact on protein?
• Frequency/rarity?
• Fit with the pedigree?
• External reports?
• Most of the time a VUS is truly uncertain
• Most VUSs will be eventually benign.
• Don’t say a VUS is an answer unless you are pretty sure.
• Sometimes a VUS really can be the answer though.
36