Bioinformatics

Slide 1 · Bioinformatics

From next-generation sequencing reads to molecular interpretation

Dr Abu Suraih Sakhri

MBBS, MD (Pathology)

www.abusuraihsakhri.com

Slide 2 · Learning objectives

Map NGS data from short reads to a clinical laboratory report

Explain alignment, variant calling, sequencing depth, and VAF

Distinguish annotations from clinical pathogenicity judgments

Relate mutational signatures, TMB, and MSI to tumor assessment

Slide 3 · Why sequencing requires bioinformatics

Sequence reads

Reference context

Variant evidence

Clinical review

NGS produces large collections of individual sequence reads

A read has little clinical meaning without genomic context

Computation organizes read-level evidence before laboratory review

Slide 4 · Sequence data are not yet a diagnosis

Data

Analysis

Annotation

Interpretation

The instrument generates DNA sequence information

Analytical pipelines transform raw observations into candidate changes

Clinical significance is established after annotation and review

Slide 5 · NGS measures many templates in parallel

Separation

Amplification

Parallel sequencing

Individual DNA molecules are physically separated

Local amplification makes each template detectable

Parallel sequencing generates one read from each amplified template

Slide 6 · How NGS differs from conventional Sanger sequencing

Feature

Sanger sequencing

Next-generation sequencing

Input

Usually homogeneous template

Heterogeneous DNA may be analyzed

Output

Composite region sequence

Many individual reads

Interpretation

Limited for mixed sequences

Computational read-level analysis

Sanger usually requires a homogeneous DNA template

Sequence mixtures can be difficult to interpret as averaged signals

Slide 7 · Spatial separation preserves molecule identity

DNA molecules

Spatially separated sites

Individual read streams

The initial process isolates templates in space

A separate site represents an individual starting molecule

Independent sites can produce separately interpretable reads

Slide 8 · Local amplification increases detectable signal

Isolated template

Local amplification

Readable signal

Amplification occurs after spatial isolation

Each original template is amplified at its own location

Enough signal is required for accurate base detection

Slide 9 · Parallel sequencing creates the data burden

Many sites

Many reads

Large data collection

Millions of templates can be interrogated simultaneously

Each amplified molecule contributes sequence information

Interpretation therefore depends on scalable computational analysis

Slide 10 · A read is a DNA fragment sequence

Reference

A

C

G

T

A

C

T

G

A

C

G

T

Read 1

A

C

G

T

A

C

Read 2

G

T

A

C

T

G

Read 3

T

A

C

T

G

A

Read 4

C

G

T

A

C

T

Original conceptual alignment schematic; Not patient sequence data.

A read contains the base sequence of one measured fragment

Short reads must be related to the reference genome

The biological question concerns the source locus, not the read alone

Slide 11 · High throughput creates analytical scale

High-throughput instruments can produce billions of reads

Data volumes reaching terabytes per day

Storage and interpretation are therefore inseparable from sequencing

Slide 12 · A pipeline is an ordered analytical workflow

Alignment

Variant calling

Annotation

Signatures

Analysis methods vary by application and specimen type

The core tasks are alignment, variant calling, and interpretation

Additional analyses can characterize aggregate mutation patterns

Slide 13 · Specimen and clinical question shape the analysis

Clinical question

Constitutional variation

Tumor alterations

Circulating tumor DNA

Constitutional disease and cancer testing ask different questions

DNA can be heterogeneous within a tissue or blood specimen

Pipeline selection and interpretation must reflect the application

Slide 14 · Read alignment provides genomic context

Reference

A

C

G

T

A

C

T

G

A

C

G

T

Read 1

A

C

G

T

A

C

Read 2

G

T

A

C

T

G

Read 3

T

A

C

T

G

A

Read 4

C

G

T

A

C

T

Original conceptual alignment schematic; Not patient sequence data.

Alignment maps sequencing reads to a reference genome

A genomic coordinate makes a read interpretable

The mapped collection becomes the substrate for subsequent comparison

Slide 15 · The reference genome is a coordinate system

Reference

A

C

G

T

A

C

T

G

A

C

G

T

Read 1

A

C

G

T

A

T

Read 2

G

T

A

T

T

G

Read 3

T

A

T

T

G

A

Read 4

C

G

T

A

C

T

Red = hypothetical alternate base; Example is not patient data.

Mapped reads can be viewed at corresponding reference positions

A reference base is the comparator for detecting variation

Alignment alone does not determine clinical significance

Slide 16 · Many reads may cover the same locus

Illustrative read coverage across neighboring positions

Site 1

Site 2

Site 3

Site 4

Site 5

Site 6

Illustrative heights, not measured depth

Multiple molecules can yield reads overlapping one site

Overlapping observations supply repeated evidence

Their accumulation provides sequencing depth at that site

Slide 17 · Alignment and variant calling are distinct

Where?

What differs?

What does it mean?

Alignment asks where each sequence fits in the reference

Variant calling asks how the sample differs at those positions

Clinical interpretation asks why a particular difference matters

Slide 18 · Ambiguous mapping illustrates a limitation

Short sequence read

Clear placement

Competing placements

Review needed

A short sequence may not uniquely identify one genomic location

Reference context and analytical review influence confidence

Not every read supports an unambiguous variant conclusion

Slide 19 · Depth is the number of reads covering a position

Illustrative read coverage across neighboring positions

Site 1

Site 2

Site 3

Site 4

Site 5

Site 6

Illustrative heights, not measured depth

Sequencing depth describes local read coverage

Depth can vary across the regions being interrogated

A reported variant must be interpreted relative to its evidence

Slide 20 · Breadth and depth answer different questions

Feature

Breadth

Depth

Question

Which regions are covered?

How many observations at a site?

Primary relevance

Scope of detection

Sensitivity to low-frequency alleles

Design influence

Target selection

Allocation of sequencing reads

Breadth describes how much genomic territory is interrogated

Depth describes how many reads cover a position

Slide 21 · Low-frequency alleles require sufficient depth

Reference and alternate molecules in a sample

R

R

A

R

R

R

R

A

R

R

R

A

R

R

R = reference support A = alternate support

A minority alternate signal can coexist with abundant reference DNA

Conceptual schematic — not patient data or an assay validation plot.

An altered allele may be present in only a minority of molecules

More overlapping reads increase the chance of observing it

Low-VAF detection therefore depends strongly on local depth

Slide 22 · Variant allele frequency is a read proportion

Reads overlapping a single locus

R

R

A

R

R

R

R

A

R

R

R

A

R

R

R = reference support A = alternate support

VAF = alternate-supporting reads / all evaluated reads at that site

Conceptual schematic — not patient data or an assay validation plot.

VAF is the fraction of reads at a locus supporting an alternate allele

VAF describes observed evidence in the sequenced specimen

It is not, by itself, a clinical classification or diagnosis

Slide 23 · VAF: Distinguish support from total coverage

Reads overlapping a single locus

R

R

A

R

R

R

R

A

R

R

R

A

R

R

R = reference support A = alternate support

VAF = alternate-supporting reads / all evaluated reads at that site

Conceptual schematic — not patient data or an assay validation plot.

The numerator is reads supporting the specified variant

The denominator is reads considered at that position

Interpret the proportion together with depth and sample context

Slide 24 · A low VAF is not the same as a small effect

Feature

Observed allele fraction

Clinical effect

Meaning

Relative read support

Potential disease relevance

Evidence

Sequencing observations

Annotation and clinical evaluation

Do not infer

Pathogenicity from frequency

Variant abundance from pathogenicity

VAF quantifies how often a change is observed in sample reads

Pathogenicity concerns biological or clinical significance

Slide 25 · Mixed cell populations complicate interpretation

Reference and alternate molecules in a sample

R

R

A

R

R

R

R

A

R

R

R

A

R

R

R = reference support A = alternate support

A minority alternate signal can coexist with abundant reference DNA

Conceptual schematic — not patient data or an assay validation plot.

NGS can analyze specimens containing different DNA molecules

Tumor-containing samples may mix altered and unaltered DNA

Observed VAF reflects the analyzed material, not a pure cell population

Slide 26 · Variant calling systematically compares sequences

Reference

A

C

G

T

A

C

T

G

A

C

G

T

Read 1

A

C

G

T

A

T

Read 2

G

T

A

T

T

G

Read 3

T

A

T

T

G

A

Read 4

C

G

T

A

C

T

Red = hypothetical alternate base; Example is not patient data.

The called result expresses a difference from a reference sequence

Calling depends on the aligned sequence observations

This step precedes biological annotation and clinical interpretation

Slide 27 · The alternate base must be supported by reads

Reference

A

C

G

T

A

C

T

G

A

C

G

T

Read 1

A

C

G

T

A

T

Read 2

G

T

A

T

T

G

Read 3

T

A

T

T

G

A

Read 4

C

G

T

A

C

T

Red = hypothetical alternate base; Example is not patient data.

A reference position provides the expected base

Aligned sequence reads may show the reference or an alternate base

The evidence needs assessment before a change is reported

Slide 28 · Variant identity and clinical meaning are separate

Observed change

Annotated change

Clinical assessment

A called sequence change is an analytical observation

Some changes are common polymorphisms

Clinical assessment determines whether a change is relevant to disease

Slide 29 · Germline polymorphisms illustrate baseline variation

Individuals naturally differ from the reference genome

Adequate sequencing is intended to identify variants within target intervals

A mismatch with the reference does not automatically imply disease

Slide 30 · Coverage constrains a negative interpretation

Feature

Positive finding

No variant detected

Evidence

Alternate reads observed

No qualifying alternate evidence

Interpretive issue

Validity and biological effect

Was detection adequately supported?

Context

Source specimen

Scope and depth of the assay

Failure to observe a variant depends on what was examined

Insufficient depth weakens inference about rare alleles

Slide 31 · Analytical detection is not pathogenicity

Detection

Annotation

Review

Report

Variant calling identifies candidate sequence alterations

Annotation adds molecular and contextual information

Laboratory interpretation assigns clinical relevance where supported

Slide 32 · Annotation turns coordinates into descriptions

Called sequence variant

Gene name

Coding / protein effect

Variant databases

Clinical context

Annotation assembles evidence; Laboratory review determines what is reportable.

A genomic change can be linked to a gene

Coding changes and predicted protein effects may be added

Population, pathogenicity, and clinical databases offer context

Slide 33 · Gene identification is a first annotation layer

Called sequence variant

Gene name

Coding / protein effect

Variant databases

Clinical context

Annotation assembles evidence; Laboratory review determines what is reportable.

The genomic location may be assigned to a named gene

Gene identity provides functional context

The gene name alone cannot establish a disease-causing effect

Slide 34 · Sequence change and protein effect differ

Feature

Coding annotation

Protein consequence

Describes

Nucleotide-level change

Possible amino-acid or functional effect

Evidence type

Sequence mapping

Computed consequence / prediction

Pitfall

Gene name is not diagnosis

Prediction is not proof

The coding-level description identifies an alteration in sequence

Protein-effect predictions describe possible consequences

Slide 35 · Population databases provide frequency context

Called sequence variant

Gene name

Coding / protein effect

Variant databases

Clinical context

Annotation assembles evidence; Laboratory review determines what is reportable.

Commonly occurring variants can be recognized through databases

Frequency information helps place a finding in perspective

Commonness and clinical significance must not be conflated

Slide 36 · Pathogenic variant resources add disease context

Called sequence variant

Gene name

Coding / protein effect

Variant databases

Clinical context

Annotation assembles evidence; Laboratory review determines what is reportable.

Databases may catalog variants recognized as pathogenic

Reported pathogenicity information requires expert assessment

The clinical laboratory remains responsible for reviewing evidence

Slide 37 · Clinical information completes the annotation

Annotated variant

Laboratory evidence

Disease context

Clinical relevance

Clinical history can affect the relevance of a molecular finding

The same observed alteration may answer different diagnostic questions

Interpretation connects the result to the patient context

Slide 38 · The laboratory review is a necessary gate

Candidate variants

Evidence review

Clinical report

Annotated output is not automatically a final clinical report

The clinical laboratory reviews identified changes

Only reviewed findings are incorporated into reporting

Slide 39 · Reports should explain biological significance

Specimen and assay context

Reviewed molecular findings

Biological significance

Clinical significance and caveats

Clinicians need more than a variant name or coordinate

A concise description should address biological meaning

The report should distinguish clinically important findings

Slide 40 · Reports should explain clinical significance

Specimen and assay context

Reviewed molecular findings

Biological significance

Clinical significance and caveats

Molecular knowledge evolves faster than routine clinical familiarity

A report translates genomic interpretation into patient-relevant terms

Evidence and remaining uncertainty should be communicated

Slide 41 · Tumor reports may contain multiple alterations

Specimen and assay context

Reviewed molecular findings

Biological significance

Clinical significance and caveats

Cancers can harbor numerous detected changes

The clinically meaningful subset requires interpretation

Variant lists alone are an insufficient summary of disease relevance

Slide 42 · Mutational signatures aggregate patterns

Individual changes

Mutation pattern

Potential process

Variant-level information

Computational synthesis

Exposure / DNA repair context

Single variants describe discrete sequence changes

Signatures examine patterns across multiple mutations

Patterns can provide clues about underlying mechanisms

Slide 43 · Some patterns suggest environmental exposure

Mutation pattern

Ultraviolet-associated

Smoking-associated

Other processes

Ultraviolet light as one signature-associated exposure

Smoking is another cited source of mutation patterns

A signature is an etiologic clue, not direct proof of an exposure

Slide 44 · DNA repair defects can shape mutational patterns

Individual changes

Mutation pattern

Potential process

Variant-level information

Computational synthesis

Exposure / DNA repair context

Failure of DNA repair alters the processes generating mutations

Informatics can detect patterns compatible with repair defects

The interpretation concerns a process, not merely a variant count

Slide 45 · Tumor mutational burden summarizes burden

Observed sequence alterations

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

Aggregate mutation burden in the assessed territory

Conceptual grouping only: No mutation count, cutoff, or specimen shown.

TMB quantifies mutation burden rather than naming one driver

Informatics may estimate this feature from NGS data

The clinical role discussed by Robbins concerns immunotherapy response

Slide 46 · Microsatellite instability signals repair dysfunction

Mismatch repair defect

Microsatellite instability

NGS assessment

Defects in mismatch repair can produce microsatellite instability

MSI is a distinct feature from the total mutation burden

NGS-derived data can be used to assess MSI status

Slide 47 · TMB and MSI are distinct readouts

Feature

TMB

MSI

Main construct

Mutation burden

Microsatellite instability

Associated process

Aggregate mutation accumulation

Mismatch repair dysfunction

Interpretive caution

Not a specific mutation

Not identical to high TMB

TMB expresses overall tumor mutation burden

MSI characterizes instability linked to mismatch repair dysfunction

Slide 48 · NGS biomarkers inform immunotherapy discussions

High TMB and MSI with checkpoint-inhibitor responsiveness

This is a clinical association, not a guarantee for an individual patient

Interpretation must remain linked to the specific clinical setting

Slide 49 · Signature analysis adds a process-level layer

Variant calls

Pattern analysis

Biological context

Variant calling lists changes at individual sites

Signature analysis considers relationships across many events

Combining layers gives a more complete tumor profile

Slide 50 · NGS scope determines what can be interpreted

Whole genome: Coding and noncoding regions

Whole exome: Coding exons

Selected gene panel

Conceptual nested scopes — area is not proportional to sequence size.

Clinical testing can target selected genes, exons, or the whole genome

Broader territory generally introduces more interpretation work

Bioinformatics must match the assay’s intended genomic scope

Slide 51 · Targeted panels prioritize relevant regions

Whole genome: Coding and noncoding regions

Whole exome: Coding exons

Selected gene panel

Conceptual nested scopes — area is not proportional to sequence size.

A panel sequences selected genes or exons

Concentrating reads can maximize depth and reduce cost

Smaller reportable territory can limit interpretation burden

Slide 52 · Whole-exome sequencing broadens coding coverage

Whole genome: Coding and noncoding regions

Whole exome: Coding exons

Selected gene panel

Conceptual nested scopes — area is not proportional to sequence size.

WES enriches protein-encoding exons

Coding exons as roughly 1.5% of the genome

The broader scope is useful in unresolved genetic disorders

Slide 53 · Whole-genome sequencing extends beyond exons

Whole genome: Coding and noncoding regions

Whole exome: Coding exons

Selected gene panel

Conceptual nested scopes — area is not proportional to sequence size.

WGS examines coding and noncoding genome territory

It can detect structural rearrangements when sequencing is sufficient

High costs and informatics challenges constrain clinical implementation

Slide 54 · The assay choice changes analytical workload

Assay

Analytical emphasis

Scope of interrogation

Gene panel

Focused depth and selected variants

Clinical gene set

WES

Broad coding-region changes

Exons

WGS

Genome-wide and structural findings

Coding + noncoding regions

Panels restrict interrogation to deliberately selected genes

WES increases coding-region interpretation

Slide 55 · Tumor profiling links alterations to care

Tumor gene panel

Molecular alterations

Prognostic information

Potential treatment relevance

Tumors contain distinct sets of somatic alterations

Gene panels seek clinically informative molecular changes

Reporting must distinguish prognostic from potentially treatable findings

Slide 56 · Constitutional disease testing seeks a genetic cause

Genetic disease assessment

Selected gene panel

WES if indicated

Interpret clinical context

NGS can investigate inherited or developmental disorders

Panels are used in disorders such as cardiomyopathy or deafness

WES can help identify molecular diagnoses after prior inconclusive testing

Slide 57 · Repeat profiling can assess acquired resistance

Initial profile

Clinical course

Repeat profile

New interpretation

A recurrent cancer may differ molecularly from the original tumor

Repeat NGS testing can identify relevant resistance-associated changes

Interpret the new findings in relation to prior treatment and specimen

Slide 58 · Liquid biopsy makes allele sensitivity important

Reference and alternate molecules in a sample

R

R

A

R

R

R

R

A

R

R

R

A

R

R

R = reference support A = alternate support

A minority alternate signal can coexist with abundant reference DNA

Conceptual schematic — not patient data or an assay validation plot.

Tumor-derived DNA may circulate in peripheral blood

Tumor-related alterations can be infrequent in mixed DNA

Low-VAF analysis and assay-specific interpretation become important

Slide 59 · Circulating DNA requires specimen-aware interpretation

Feature

Tumor tissue analysis

Circulating tumor DNA analysis

Material

Direct tumor-containing specimen

DNA recovered from blood

Evidence

Tumor sequence profile

Tumor-associated signal in circulation

Caution

Tumor heterogeneity

Low-frequency signal and sample context

Circulating tumor cells and cell-free tumor DNA are different analytes

NGS can detect tumor-associated sequence changes in blood

Slide 60 · A defensible report answers three questions

Specimen and assay context

Reviewed molecular findings

Biological significance

Clinical significance and caveats

What sequence change was detected, and with what supporting evidence?

What is its relevant molecular or biological annotation?

What clinical interpretation is justified for this specimen?

Slide 61 · Case: A suspected inherited disorder

Which step turns a called variant into a useful diagnostic result?

Interpretation integrates annotation and the clinical phenotype; A reference mismatch alone is insufficient.

Hypothetical teaching case; No patient-derived results are shown.

A child with a suspected genetic disorder undergoes sequencing

Many sequence differences are identified relative to the reference

Prioritize laboratory review, annotation, and phenotype concordance

Slide 62 · Case: Heterogeneous cancer sequencing

Does a low observed VAF make the alteration clinically unimportant?

No. VAF describes the observed fraction of reads, not the variant’s biological or clinical consequence.

Hypothetical teaching case; No patient-derived results are shown.

A tumor specimen contains altered and unaltered DNA molecules

A low-frequency alteration is seen among overlapping reads

Depth, specimen composition, and reportable clinical meaning must be assessed

Slide 63 · Case: Tumor pattern information

Why does a mutational signature add information beyond a variant list?

It summarizes a pattern that may indicate an exposure or DNA repair mechanism.

Hypothetical teaching case; No patient-derived results are shown.

A cancer profile shows numerous sequence alterations

Aggregate analysis identifies a pattern relevant to a repair process

Individual variants and process-level features provide complementary evidence

Slide 64 · The principal pitfalls are category errors

Feature

Analytical output

Question answered

Alignment

Where does a read map?

Coordinates

Variant call / VAF

What differs / in what fraction?

Local evidence

Annotation / review

What does the finding mean?

Biology + clinical context

Alignment is not variant calling; Variant calling is not diagnosis

Sequencing depth is not the same as genomic breadth

Slide 65 · An integrated molecular-diagnostic workflow

Reads

Alignment

Variant calls

Annotation

Report

The reference genome anchors the aligned reads

Read evidence becomes candidate variants and annotated findings

Clinical laboratory review converts selected findings into a report

Slide 66 · Knowledge check: Interpret the analytical layers

A low-VAF finding needs assessment of read support and context

An annotated coding change still requires clinical interpretation

A tumor mutation pattern can provide mechanism-level information

Slide 67 · References and source boundary

Robbins, Cotran & Kumar: Pathologic Basis of Disease, 11th ed. (2026)

Chapter 5, “Genes and Human Diseases,” Bioinformatics, pp. 164–165

Adjacent NGS principles and clinical applications provide teaching context

Schematics are original conceptual illustrations, not patient observations