Clinical Variant Interpretation with the Integrative Genomics Viewer (IGV) for Molecular Pathologists

Summary

The integrative genomics viewer (IGV) software offers a rich, intuitive window into the underlying data for clinical variant calling. Here we present five vignettes that demonstrate how to perform swift differentiation between genuine pathogenic events and artifacts, ultimately reinforcing diagnostic confidence for practitioners engaged in molecular sign-out of next-generation sequencing assays.

Abstract

The integrative genomics viewer (IGV) is a pivotal tool in clinical genomics, enabling the visualization and interpretation of complex sequencing data. Bringing clinical knowledge to bear with visual evaluation of sequencing results is the primary means by which molecular pathologists and other professionals assess and finalize cases. A variety of software tools can assist, but their relationship to the underlying data must be understood and applied systematically. This study includes essential background on next-generation sequencing (NGS) data file types (e.g., FASTQ, BAM, VCF) with a discussion of their format and purpose. We then describe features of IGV that derive nuances from these files. We utilize a series of curated practical cases based on clinical vignettes through which the reader will interact with clinical NGS sequencing data using the IGV software to review various types of clinically relevant variants relative to the human reference genome. These clinical vignettes have been curated to describe examples of some of the complexities of interpretation of genomic data, and how utilizing IGV as part of a routine workflow can provide additional interpretive information for variants beyond routine bioinformatic software algorithm variant calls. The visual inspection of genomic variants utilizing the tools within IGV can unmask subtle contextual cues (i.e., variant allele frequency, strand bias, tissue-specific context) that can influence the interpretation of genomic variants. Although this study focuses on using IGV for the detection and interpretation of somatic variants, the provided applications can be extrapolated for use in the germline setting, including analysis of complex variants and detection of mosaicism.

Introduction

Oncology care has been transformed in recent years by the use of next-generation sequencing (NGS) technologies to inform diagnosis, prognosis, and therapy selection¹. NGS testing has become the standard of care in multiple tumor types for 1) establishing a diagnosis of a molecularly defined tumor, 2) using tumor-type agnostic biomarker results for therapy decisions, and 3) determining risk stratification for specific tumor types^2,3. Accurate and timely results from NGS and other biomarker testing have become a critical component for patient care and planning in oncology.

NGS testing is typically performed in either large reference laboratories or in hospital-based/academic laboratories, when institutions have sufficient expertise and resources to develop and maintain such testing. The Genomics Organization for Academic Laboratories (GOAL) organization was conceived as a collaborative network, focused on the sharing of academic knowledge and reagent costs⁴. A significant effort of the GOAL consortium has involved training sessions for critical components for the review and reporting of clinical studies. These have been developed as practical instructional sessions and include an interactive workshop on the integrative genomics viewer (IGV) to demonstrate its basic and advanced features in genomic medicine.

IGV is a data visualization tool designed to allow users to easily review NGS data and other large datasets, given that most NGS data are vast and unwieldy without a graphical user interface⁵. IGV was not initially designed for clinical variant analysis. However, given its ease of use for all molecular professionals, regardless of their level of bioinformatics experience, it has been widely adopted by many laboratories as part of the complex data review workflow⁶. This study was assembled as a mechanism to disseminate this information to the broader community of genomic laboratory professionals.

The clinical need for using IGV as part of a clinical workflow includes detecting sequencing errors, handling complexities in genomic variant calling, and other recurrent alterations where additional visualization can assist in appropriate data interpretation⁷. The use of IGV requires standard bioinformatic files originating from an NGS study⁸. Core features of the IGV software allow visualization of sequencing data and require configuring a set of preferences and options for clinical variant interpretation (Supplementary File 1).

The short-read sequencing workflow is typified by FASTQ, binary alignment map (BAM), and variant call format (VCF) file standards. NGS achieves its high throughput by generating millions of short reads in parallel. While this strategy greatly accelerates data production, each read is only a fragment of the original DNA or RNA. It is affixed to some position on an NGS flow cell by its adapter sequence. Sequence and base quality information from optical processing is recorded in FASTQ files. These raw reads are nucleotide sequences along with their per-base quality scores. Therefore, an early step in processing raw reads is to align each read to a reference genome⁸. As each read is assigned a specific location in the reference genome with the highest probability of alignment, the reads that correspond to the same location "pile up" at each locus, and such data is stored in a BAM file following the sequence alignment mapping (SAM) specifications (available at https://samtools.github.io/hts-specs/SAMv1.pdf; last verified: 9/30/2025) (Figure 1A). This information can be visually inspected by a genome browser such as IGV. As illustrated in Figure 1B, comparing the reads to one another or to the reference sequence itself can indicate differences or discrepancies between the nucleotide sequences. Summaries of only these "variant" positions across the entire genome are compiled into a VCF file, which can therefore be much smaller in size than a BAM file but still contain much of the useful information required for variant review. For that reason, the VCF file is commonly the final output of variant-calling pipelines (Figure 1C). Thus, while FASTQ files contain the raw read and quality data, the BAM file provides insight into how confidently those reads map to the genome, and the VCF represents the differences between the reference sequence and the sample sequence, providing a much more manageable amount of data for review for potential reporting.

The six clinical vignettes in this study illustrate common analytical scenarios associated with somatic variant detection in molecular laboratories as well as the problems that complicate the correct understanding and interpretation of the underlying data. Overall, these vignettes were chosen to illustrate increasingly sophisticated interpretative scenarios and how utilizing advanced features in IGV can better describe comprehensive genomic features. Although this study focuses on somatic variant interpretation, these features are applicable for utilizing IGV in the visualization and interpretation of germline alterations⁶. Of note, the terms used in this article are based on sequencing-by-synthesis chemistry and, where appropriate, other analogous terms/concepts for other platforms.

Each clinical vignette begins with a very brief clinical context, followed by instructions on how to investigate the variant or variants and resolve the question: which (if any) variant present is real and which (if any) is an artifact. The clinical context, along with the particulars of the sequencing data, is important for answering that question.

Protocol

The data presented here are a mixture of simulated and real patient NGS data. However, all patient data is de-identified through a mixture of label stripping and single-locus read extraction. This protocol follows the applicable guidelines for the human research ethics committee at all of the authors' institutions.

1. Downloading the most recent instance of Integrative Genomics Viewer (IGV)

NOTE: IGV is a free downloadable program for visualizing various genomes, including the human genome (https://igv.org/doc/desktop/#DownloadPage/). IGV is periodically updated to include new versions of transcripts, features, and tracks, with the most current version typically at the top of the list.

1. Download the file available from the uniform resource locator (URL) above, named 'IGV for Windows, Java Included,' to the desktop, which may require administrative credentials, depending on the institution.
2. Choose the human genome build that is planned to be called from the reports (e.g., hg19 or hg38), which are hosted within IGV.
   NOTE: This study used IGV 2.19.7. In this protocol, the version for Windows, with Java included, was downloaded. IGV was run with the default memory allocation of 8Gb of RAM. The settings recommended within the IGV preferences tab and other sections in IGV are annotated in Supplementary File 1.

2. Understanding the clinical vignettes and downloading the necessary files

1. Navigate to the GitHub repository: All files are available for download in the GitHub repository https://github.com/Eitan177/Demo_IGV.
   NOTE: There is a session file in the GitHub repository for each vignette, and loading this session file obviates the need to load files separately and navigate to the locus in question because the session file handles these steps.
2. Load a session file via the file menu. Click on File > Open Session, and select the appropriate sessions file.

3. Setting up the coverage track

NOTE: In IGV, the main window includes a "coverage track". In certain contexts, even a small fraction of reads containing specific cancer variants may be clinically significant. Because of this, achieving very high read depth across the portions of the genome being assayed is crucial. If only 2% of reads contain a given variant, a read depth of hundreds or thousands may be required to convincingly call the variant – hence the term "deep sequencing".

1. Ensure coverage track is present: From the overhead menu, click View > Preferences, and select the Alignments tab. At the top of the tab, there is a box labeled 'Show coverage track'. Ensure it is checked.
2. Ensure a clinically significant low-level variant threshold: Navigate to the Alignments tab as detailed in step 3.1 and scroll down until the box labeled Coverage allele-fraction is visible. Set the number in the box to 0.01 and hit Save (the preferences window should close upon hitting Save).

4. Setting up soft-clipped reads

NOTE: This setting is invaluable for detecting larger structural variants, such as gene rearrangements, copy number changes, or combined insertion–deletion events. The advanced vignettes—such as those focusing on EGFR amplifications and deletions—demonstrate how soft-clipped reads can corroborate the presence of structural variants that might otherwise be missed by standard variant calling methods.

1. Enable soft-clips: From the overhead menu, click View > Preferences, and select the Alignments tab. Scroll down to the box Show soft-clipped bases and ensure that it is checked (Supplementary File 1) to reveal partially-aligned reads—those whose edges do not map cleanly to the reference and are "clipped" off.

5. Clinical vignettes

1. Vignette #1: Distinguish a true KRAS (Kirsten rat sarcoma virus oncogene homolog) pathogenic variant from a sequencing artifact.
   NOTE: Case A involves a 65-year-old male presenting with a newly diagnosed lung adenocarcinoma, where histologic assessment estimated approximately 80% tumor. NGS profiling identified a KRAS NM_004985.5 c.34G>T (p.G12C) variant, which is a known driver mutation in non-small cell lung cancer (NSCLC). The resulting missense variant KRAS c.34G>T (p.G12C) occurs in ~13% of lung adenocarcinomas and confers eligibility for targeted therapies⁹. Case B involves a 54-year-old female with a known sarcoma (estimated 50% tumor and 50% normal tissue). The identical KRAS c.34G>T (p.G12C) variant was identified in this sample. However, because KRAS c.34G>T (p.G12C) is seldom reported in sarcoma, its presence prompts suspicion of a possible sequencing artifact. 
   1. IGV workflow (Session file KRAS.xml)             
      1. Load data: Select Hg19 in the genome drop-down and load the bam file, kras2A_fixed_backbone_2A_new_final_sorted.bam, and navigate to chr12:25,398,281–25,403,833 to focus on the KRAS gene region. Note the index file (same name as the bam file, but ending in .bai) needs to be in the same directory as the bam file.  
      2. Investigate coverage and VAF: Review codon 12 in Exon 2 (NM_004985.5) and click the variant position in the coverage track to show the read depth at that position. Note a G→T substitution (transversion event), with 35%–40% VAF, highly consistent with a heterozygous, clonal driver mutation (Figure 2A), particularly in the context of approximately 80% tumor cellularity.   
      3. Look for StrandBias: Color-code by strand (see Supplementary File 1, Step 4) to reveal a near 50:50 forward (red): reverse (blue) distribution in the lung sample.  
      4. Repeat steps 5.1.1.1–5.1.1.3 for case B, the sarcoma sample: Load the bam file, kras2B_fixed_backbone_2B_new_final_sorted.bam, and note the <1% VAF and near exclusive forward strand support, indicating a PCR/alignment artifact (Figure 2B).
2. Vignette #2: Resolve a multi-nucleotide variant (MNV) misclassification.
   NOTE: A 50-year-old female with metastatic melanoma undergoes NGS-based tumor profiling. Analysis identifies a potential BRAF (B-Raf proto-oncogene, serine/threonine kinase) c.1798_1799delinsAA mutation. While BRAF NM_004333.6: c.1799T>A (p.V600E) is more common across various solid tumors, BRAF c.1798_1799delinsAA (p.V600K) is relatively specific for melanoma, as UV-light can induce C > T transitions within dipyrimidines¹⁰. According to HGVS rules, two single-nucleotide variants (SNVs) in cis should be named as a single MNV. In many cases, a bioinformatics pipeline may incorrectly identify this complex variant as two adjacent SNVs in cis: BRAF c.1798G>A and BRAF c.1799T>A. Because there are technical (i.e., variant caller-related) and biological (cis vs trans distinction) considerations, there is concern that the mutation could be reported incorrectly at the time of sign-out unless there is visualization of the data during clinical review.      
   1. IGV workflow (Session file BRAF.xml)           
      1. Load data: Select Hg38 in the genome drop-down and load the bam file, kras2B_fixed_backbone_2B_new_final_sorted.bam, and navigate to chr7:140,453,136–140,753,336. Note the index file (same name as the bam file, but ending in .bai) needs to be in the same directory as the bam file.
      2. Zoom in on codon 600 of BRAF: Note a GT→AA dinucleotide change based on adjacent coverage track positions that are green and blue rather than gray, consistent with p.V600K. 
      3. Scan for two individual SNVs to determine if they are in cis or trans: Note that both SNVs are mapped to the same reads and are in cis.  
      4. Investigate strand and coverage: Color-code reads to show roughly even forward/reverse support (see Supplementary File 1, Step 4), and a VAF exceeding 50% consistent with a clonal driver. A coverage track threshold of 0.01 (see Supplementary File 1, Step 3 for setup) ensures that partial subclones (if present) will remain visible (Figure 2C).  
      5. Note context and MNV nomenclature: The two nucleotide changes are adjacent to each other. Based on the expert consensus, call SNVs within 50 bp distance that map on the same sequence read as a single "complex" or "delins" variant¹¹. 
3. Vignette #3: Investigate a KIT (KIT proto-oncogene, receptor tyrosine kinase) exon 9 insertion in a gastrointestinal stromal tumor (GIST)
   NOTE: A 58-year-old male patient with a gastrointestinal stromal tumor undergoes resection, and NGS testing is ordered. Results show a possible variant in KIT. KIT mutations are a frequent driver in GISTs, occurring in approximately 75% of cases, and are targetable. After exon 11 activating mutations, KIT exon 9 activating mutations are the most common drivers identified, though exon 9 mutated GISTs and exon 11 mutated GISTs behave differently with respect to tyrosine kinase inhibitor sensitivity¹². 
   1. IGV workflow (Session file KIT.xml)
      1. Load data: Select the Hg19 genome and load the BAM file, KITdupmod2.bam and make sure the index file (same name as the bam, though ending in .bai) is in the same directory. Navigate to chr4:55,592,144-55,592,257, corresponding to a portion of exon 9 of the KIT gene.
      2. Squish the view: Right-click the alignment track and select Squished to condense the view (Figure 3A).
      3. Locate the insert: Identify the insertion marker, shaped like an arrowhead, immediately above the coverage track, located between positions 55,592,178 and 55,592,179, indicating a short insert (Figure 3B,C).
      4. Interact with the insert: Click the arrowhead to expand the view and click on one of the reads with bases filling the gap to view the details in a pop-up. Compare the inserted sequence with the reference sequence to determine if it is a duplication (Figure 3D,E).
      5. Look at the coverage: Click on the coverage track immediately adjacent to the insert arrowhead to display the number of supporting reads, which can be compared to the total read count to determine the approximate variant allele frequency (VAF), in this case ~28%.
      6. Check for strand bias: Sort reads by read strand and view soft-clipped bases (see Supplementary File 1, section 3) to show a region of interest adjacent to the insertion in the reverse reads (Figure 3F).
4. Vignette #4: Investigate an EGFR (epidermal growth factor receptor) Exon 19 deletion in non-small-cell-lung cancer (NSCLC).
   NOTE: A 70-year-old female with newly diagnosed NSCLC undergoes surgical resection of the mass. NGS is ordered on a section of tissue containing sufficient tumor purity, and results show a possible variant in EGFR. Many EGFR mutations are susceptible to tyrosine kinase inhibitors, and this variant may have an impact on therapy and prognosis for the patient¹³.
   1. IGV workflow (Session file EGFR.xml)
      1. Load data: Select the Hg19 genome, load the bam file EGFRE746_A750delmod.bam, and make sure the index file (same name as the bam, though ending in .bai) is in the same directory, and navigate to chr7:55,242,420-55,242,513, corresponding to a portion of exon 19 of the EGFR gene.
      2. Investigate deletion using the coverage track: Review the coverage track to identify a small region of lower coverage not positioned at the ends of the reads (Figure 4A,B). Examination of the coverage track shows lower coverage over a small portion of exon 19.
      3. Investigate deletion using the reads: Expand the alignment view to show reads with an alignment gap, represented as a black line within the read (Figure 4C,D).
      4. Consider genomic context: Investigate the nucleotide sequence of the reference involved in and in the immediate surrounding of the deletion. Assess the deletion call in the context of the position within the reference sequence, as deletions in homopolymer or repetitive regions may represent a sequencing artifact, particularly when present at low variant allele frequency. See the Viewing a small deletion portion of Supplementary File 2 for a detailed description of the workflow.
5. Vignette #5: Investigate EGFR alterations in a patient with glioblastoma.
   NOTE: A 77-year-old female with IDH wild-type glioblastoma undergoes next-generation sequencing (NGS) testing. Her results indicate a potential EGFR copy number gain and a potential EGFRvIII rearrangement (deletion of exons 2–7). Both the EGFR amplification and the EGFRvIII alterations have prognostic and therapeutic significance in glioblastoma. Additionally, these two alterations frequently co-occur, meaning the presence of one may support the validity of the other when detected at low levels.
   1. IGV workflow (Session file copynumber.xml):
      1. Load data: Select the Hg38 genome, load the segment file, proband.seg, and comparator samples, c1.seg-c9.seg, by dragging and dropping them into the IGV window. Navigate to Chr7.
         NOTE: The segmentation files generated from CNVKit are provided for downloading. The full workflow for using CNVKit to generate copy number quantification is outside the scope of this study. However, briefly, CNVKit version 0.9.9 (available for download from https://github.com/etal/cnvkit) with a premade pooled reference consisting of normal, euploid specimens was used. The input consists of a bed file containing 9,144 CNV backbone probe locations, distributed across the genome. An important consideration here is that the specifics of the copy number pipeline inputs and parameters are less important than the fact that the samples are processed using a setup that is the same for all samples on a run. This facilitates recognition of systemic bias.
      2. Evaluate the heatmap: Compare relative color intensity between the sample of interest (locate the row of the heatmap corresponding to this sample, labeled proband) and other samples (c1-9) to distinguish true copy number changes from artifacts.
         NOTE: Vertical visual comparison of a locus across all samples is the means to identify anomalies (qualitative differences based on color differences cross rows rather than quantitative differences) at a given locus in each sample relative to the background. Refer to Figure 5 and Viewing copy number changes section of Supplementary File 2 for detailed steps.
6. Vignette #6: Investigate EGFRvIII in a patient with glioblastoma.
   NOTE: The same patient also has evidence of an EGFRvIII rearrangement (deletion of exons 2–7). Both the EGFR amplification and the EGFRvIII alterations have prognostic and therapeutic significance in glioblastoma. Additionally, these two alterations frequently co-occur, meaning the presence of one may support the validity of the other when detected at low levels.
   1. IGV workflow (Session file egfr_vIII.xml):
      1. Load data: Select the Hg19 genome, load the patient's bam file, proband.bam, and make sure the index file (same name as the bam, though ending in .bai) is in the same directory, and navigate to the EGFR gene.
      2. Visualize breakpoints of EGFRvIII: Adjust view to examine candidate breakpoints individually (upstream of exon 2 and downstream of exon 7 for EGFRvIII), and perform 'sort-by-base'  (or use shortcut Ctrl-s on Windows/Linux or Cmd-s on Mac) at the breakpoints to visualize pileups of soft-clipped reads.
      3. BLAT align reads: Right-click a soft-clipped sequence and select BLAT sequence (IGV uses defaults from UCSC, see https://ucsc.crg.eu/FAQ/FAQblat.html for BLAT default parameters). Click the first row and observe the Blat track that appears at the bottom of the IGV track section.
         NOTE: Even a single read that aligns to the junction of exon 1/intron 1 and to the junction of intron 6/exon 7 may be sufficient evidence of EGFRvIII, as long as cross-contamination from another sample is ruled out as a reason for this read. Refer to Figure 6 and the Viewing structural rearrangement section of Supplementary File 2 for detailed steps.

Results

Of note, this study focuses on describing the nuances of variant review in IGV for clinical reporting. The final clinical reporting should be dependent on each individual laboratory's validated pipeline, quality criteria, and reporting policy. A complete discussion of this important topic is beyond the scope of the current study; however, recommended standards and guidelines have been put together by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology⁷ and are regularly implemented in clinical laboratories.

Clinical Vignette #1 interpretation:

Given the high VAF, balanced strand distribution, and known incidence of KRAS c.34G>T (p.G12C) in lung cancer (Figure 2A), the variant in case A can result in targeted therapy and, as such, is crucial to report clinically. Despite a positive call for the same KRAS c.34 G>T variant in the second (sarcoma) case (Figure 2B), multiple warning signs of a false-positive call are present: (1) the reported variant allele fraction (VAF) is ~1%, which is vastly below what a driver event would be expected to be in the setting of a sample estimated to have even a low tumor cellularity (20%) and (2) the reads supporting this variant show strand bias, appearing almost exclusively on the forward-strand. Together with a lack of biological plausibility for this type of cancer, the evidence suggests that the finding in case B is an artifact.

Clinical Vignette #2 interpretation:

This vignette illustrates multiple distinct errors about which clinicians must be aware. First, variant callers may appropriately or inappropriately call adjacent genetic variants either as two SNVs or as a single MNV (Figure 2C). This is due to the difference in the implementation of haplotype-aware variant calling algorithm in somatic variant callers. Only a few, more recently published variant callers, incorporate this feature^14,15. Second, technical and nomenclature-based variations such as these are fed into external databases by sequencing centers using different bioinformatic pipelines. They therefore may contain multiple distinct accounts of what ought to be listed as a single genetic variant. Unless database phasing occurs periodically, this can have a variety of untoward effects for clinical interpretation downstream. Taken together, these possible sources of error can conspire to lead to an incorrect patient report. This complex scenario highlights the value of variant visualization using high-quality software such as IGV.

Clinical Vignette #3 interpretation:

Given the multi-read support for the duplication and known mechanism of constitutive activation of KIT with exon 9 duplications in GIST, we can conclude that this KIT c.1504_1509dup (p.A502_Y503dup) variant is real and not an artifact (Figure 3). KIT mutations are frequent drivers in GISTs, occurring in approximately 90% of cases. Accurate variant characterization is critical here because while exon 11 activating mutations are the most common, exon 9 mutated GISTs behave differently with respect to tyrosine kinase inhibitor sensitivity, thus altering the therapeutic approach. Given the clinical importance of KIT mutations in GISTs, this variant should be included in the final report.

Clinical Vignette #4 interpretation:

Given the coverage track, stereotypical short deletion appearance, and read support, and the known mechanism of constitutive activation of EGFR exon 19 deletions in NSCLC, we can conclude that this EGFR c.2235_2249del (p.E746_A750del) variant is real and not an artifact (Figure 4). Many EGFR mutations, including this well-known exon 19 deletion, are highly susceptible to tyrosine kinase inhibitors. Given the clinical importance of EGFR mutations in NSCLCs, this variant should be included in the final report.

Clinical Vignette #5 and Vignette #6 interpretation:

The heatmap views confirm the EGFR amplification and the breakpoint junction between exons 1 and 8, evaluated with BLAT, confirm both findings in this patient (Figure 5 and Figure 6). EGFR amplification is generally considered a prognostic factor of more aggressive disease, and EGFRvIII gene rearrangement is a GBM-specific biomarker and is potentially targetable^16,17.

Selected workflows may be reproduced using files saved to the main branch of the GitHub repository https://github.com/Eitan177/Demo_IGV (updated 6/9/2026), which contains six session files, one for each vignette, five pairs of bam and bai files, one for each vignette using an alignment file, and 10 seg files for vignette five, which uses segment files rather than alignment files.

Figure 1: NGS file standards and sequencing workflow. (A) Short reads generated on a flow cell are tracked by optical measurements and stored in FASTQ format, which includes both sequence data and per‐base quality scores. These reads are then sorted and aligned to a reference genome, yielding a BAM file with each read "piled up" at its matching locus. (B) Discrepancies between the reference and the reads (e.g., point mutations, small indels) are identified by variant‐calling algorithms. (C) Finally, these detected variants are compiled into a VCF file, summarizing their positions, reference/alternate bases, and other annotation data. Please click here to view a larger version of this figure.

Figure 2: IGV‐based visualization of KRAS c.34G>T (p.G12C) and BRAF c.1798_1799delinsAA (p.V600K). (A) Visualizing a real variant (KRAS c.34G>T (p.G12C)) in IGV, demonstrating a high read depth (thousands of reads) and ~35–40% of reads carrying a G→T change. Forward (red) and reverse (blue) reads are balanced, suggesting a true heterozygous mutation. (B) Suspected artifact in non‐relevant tissue showing a variant with minimal read support (1% VAF). Most variant reads appear on the forward strand (red), indicating strand bias. (C) A complex variant BRAF c.1798_1799delinsAA (p.V600K) seen in IGV revealing a two‐base substitution (GT→AA) in cis. Please click here to view a larger version of this figure.

Figure 3: Visualizing a small insertion. (A) "Squished" view of KIT exon 9 in IGV helps to identify the region of interest. (B) Region of interest (red box) identified by visualizing the insertion bar and soft clipped bases. (C) Setting the "Expanded" view in IGV to allow visualization of the insertion. (D) Clicking on the insertion bar (red box) displays the inserted sequence. (E) The inserted sequence is compared to the reference sequence to identify the duplicated bases. (F) Sorting the reads by read-strand, showing that soft clipped bases correspond to the insertion in the reverse reads. Please click here to view a larger version of this figure.

Figure 4: Visualizing a small deletion. (A) "Squished" view of EGFR exon 19 showing the deletion present in numerous reads. (B) Using the coverage track to review the number of reads at positions within and adjacent to the deletion. (C) "Expanded" view showing the deletion, identifying the number of deleted bases. (D) Comparison with the reference sequence (red box) identifies the deleted bases. Please click here to view a larger version of this figure.

Figure 5: Visualizing a copy number gain. (A) There are six columns in a seg file: sample ID, chromosome, start, end, markers in the segment, and mean of the segment. (B) Loading the seg files directly into IGV allows for convenient visualization. (C) Visualization of the data in a genome-wide view. Each sample will appear as a row of data, with colors specifying values, red for >0 and blue for <0, with more intense colors reflecting values farther from 0. Upon opening the sample of interest and nine random samples from the same run, the results in a multi-sample heatmap with ten rows. The sample of interest is highlighted at the bottom in the red box. (D) Narrowing the view to the EGFR locus, a red bar is seen in the sample of interest at the bottom, and white or blue bars for the random samples. This color differential between the sample of interest and the random samples in other rows demonstrates that the amplification call in the sample of interest is above background or noise at this locus. (E) Zooming out to chr7, a contrast is noted between the locus containing EGFR in our sample and not in the random samples. The EGFR amplification displays as a red sliver in the bottom sample. (F) Narrowing the view to the MET locus, MET amplification is replicated to varying degrees across the nine random samples along with the sample of interest, suggesting this amplification call is an artifact. This highlights the difference in appearance between a true copy number call at the EGFR locus and an artifact. Please click here to view a larger version of this figure.

Figure 6: Visualizing an EGFRvIII gene rearrangement. Prior to working through this case, ensure that visualization of soft-clipped bases has been set up (See Supplementary File 1 for additional details). (A) Navigate to the first candidate junction and set the view of the alignment track to "Squished". (B) Move across the locus, iteratively applying the sort-by-base shortcut, "control-s", until a pileup of similarly soft-clipped reads becomes visible at the top of the alignment track. The greater quantity of similarly soft-clipped reads (red box) in the pileup, the greater the confidence of a structural rearrangement. (C) The BLAT tool pulls up the sequence of the soft-clipped bases and identifies where in the genome that sequence is present along with a quantitative score measuring how closely the sequence aligns with a specific sequence in a target database. Selection of a particular row will navigate to that location. (D) Using the "control-s" sort by base function again demonstrates a soft-clip pileup, this time on the opposite side of the reads. To confirm that there are no additional bases in the sample with respect to the hg38 reference, right-click a read with a run of soft-clipped bases and again BLAT the soft-clip bases. The ensuing table displays the first breakpoint of the junction, the locus initially visualized, the exon 1-intron 1 junction. (E) Selection of the top row returns to the first breakpoint. The bases at the junction of the BLAT reference alignment are identical to the soft-clipped bases used as the query, indicating that the entire run of soft-clipped reads is contained in the reference sequence alignment. Please click here to view a larger version of this figure.

Supplementary File 1: Recommended IGV setup.Please click here to download this file.

Supplementary File 2: Condensed workflows.Please click here to download this file.

Discussion

Accurate variant review for clinical reporting is dependent upon complex bioinformatic algorithms and the incorporation of visual review of complex variants utilizing IGV^5,6. This study was designed to compile a concise resource of recommended settings to visualize simple and complex genomic alterations in IGV (Supplementary File 1). This study focuses on common somatic alterations and how to review them in IGV.

The protocol begins with assessing a simple substitution mutation (Figure 2A) and a complex substitution mutation (Figure 2C) with particular attention to determining whether the identified finding is real versus an artifact. It then moves on to a simple insertion (Figure 3) and simple deletion (Figure 4), highlighting how these alterations look in IGV. Lastly, it discusses complex genomic alterations, including copy number alterations (Figure 5) and structural rearrangements (Figure 6). Condensed workflows highlighting the key steps for each of these features are summarized in Supplementary File 2. This study aims to give the reader familiarity with how different genomic alterations appear in IGV⁵.

Of note, IGV is not utilized as a pipeline tool to independently detect variants. Although outside the scope of this manuscript, laboratory policies and procedures should address reportable variants based on assay validation limitations for variant allele fraction and depth of coverage, whether orthogonal confirmation is required for variants under certain circumstances, complexities involving bioinformatics software versioning, and interobserver variation in visual interpretation are strongly recommended for research laboratories and required for clinical laboratories^1,7.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
BLAT	BLAT Search Genome	https://genome.ucsc.edu/cgi-bin/hgBlat	Standalone BLAT webpage. BLAT API is available within IGV, no additional installation required
CNVKit	GitHub	https://github.com/etal/cnvkit	Public gitHub repository with CNVKit free for download. Instructions for downloading/running CNVKit are available here as well.
Comparator segment file #1 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c1.seg	Copy Number Files
Comparator segment file #2 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c2.seg	Copy Number Files
Comparator segment file #3 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c4.seg	Copy Number Files
Comparator segment file #4 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c5.seg	Copy Number Files
Comparator segment file #5 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c6.seg	Copy Number Files
Comparator segment file #6 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c7.seg	Copy Number Files
Comparator segment file #7 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c8.seg	Copy Number Files
Comparator segment file #8 for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c9.seg	Copy Number Files
Comparator segment file for vignettte #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/c3.seg	Copy Number Files
Human Genome Sequence	Human Genome Sequence	https://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/latest/hg38.fa.gz	Downloadable fasta file containing the Hg38 human genome build.  This is available within IGV, no additional download is required
Human Genome Sequence	Human Genome Sequence	https://hgdownload.soe.ucsc.edu/goldenPath/hg19/bigZips/latest/hg19.fa.gz	Downloadable fasta file containing the Hg19 human genome build. This is available within IGV, no additional download is required
IGV Software bundle	IGV Software	https://data.broadinstitute.org/igv/projects/downloads/2.19/IGV_Win_2.19.7-WithJava-installer.exe	IGV software version 2.19.7 with Java (JDK 21), Free for Download. The link here is for windows though links for Linux and Mac are also available.
Segment file for vignettte #5 case	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/proband.seg	Copy Number Files
Tutorial GitHub Repository	GitHub	https://github.com/Eitan177/Demo_IGV/tree/main	Public GitHub repository with all necessary files to reproduce the six vignettes
Vignette #1	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/KRAS.xml	Session Files
Vignette #1 Case A bai index	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/kras2A_fixed_backbone_ 2A_new_final_sorted.bam.bai	Alignment Files
Vignette #1 Case A bam alignment file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/kras2A_fixed_backbone_ 2A_new_final_sorted.bam	Alignment Files
Vignette #1 Case B bai index	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/kras2B_fixed_backbone_ 2B_new_final_sorted.bam.bai	Alignment Files
Vignette #1 Case B bam alignment file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/kras2B_fixed_backbone_ 2B_new_final_sorted.bam	Alignment Files
Vignette #2	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/BRAF.xml	Session Files
Vignette #2  bai index	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/fig2C_braf_v600k_2bp_ backbone_2CD_new_final_sorted.bam.bai	Alignment Files
Vignette #2  bam alignment file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/fig2C_braf_v600k_2bp_ backbone_2CD_new_final_sorted.bam	Alignment Files
Vignette #3	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/KIT.xml	Session Files
Vignette #3 bai index file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/KITdupmod2.bai	Alignment Files
Vignette #3 bam alignment file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/KITdupmod2.bam	Alignment Files
Vignette #4	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/EGFR.xml	Session Files
Vignette #4 alignment file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/EGFRE746_A750delmod.bam	Alignment Files
Vignette #4 bai index	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/EGFRE746_A750delmod.bai	Alignment Files
Vignette #5	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/copynumber.xml	Session Files
Vignette #6	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/egfr_vIII.xml	Session Files
Vignette #6 bai index	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/proband.bai	Alignment Files
Vignette #6 bam alignment file	GitHub	https://github.com/Eitan177/Demo_IGV/blob/main/proband.bam	Alignment Files