Accurate and Automated Genotyping of the CFTR Poly-T/TG Tract with CFTR-TIPS

Abstract

Cystic fibrosis is caused by biallelic pathogenic variants in the CFTR gene, which contains a polymorphic (TG)_mT_n sequence (the “poly-T/TG tract”) in intron 9. While T₉ and T₇ alleles are benign, T₅ alleles with longer TG repeats, e.g., (TG)₁₂T₅ and (TG)₁₃T₅, are clinically significant. Thus, professional medical societies currently recommend reporting the TG repeat size when T₅ is detected. Sanger sequencing is a cost-effective method of genotyping the (TG)_mT_n tract; however, its polymorphic length substantially complicates data analysis. We developed CFTR-TIPS, a freely available web-based software tool that infers the (TG)_mT_n genotype from Sanger sequencing data. This tool detects the (TG)_mT_n tract in the chromatograms, quantifies goodness of fit with expected patterns, and visualizes the results in a graphical user interface. It is broadly compatible with any Sanger chromatogram that contains the (TG)_mT_n tract ± 15 bp. We evaluated CFTR-TIPS using 835 clinical samples previously analyzed in a CLIA-certified, CAP-accredited laboratory. When operated fully automatically, CFTR-TIPS achieved 99.8% concordance with our clinically validated manual workflow, while generally taking less than 10 s per sample. There were two discordant samples: one due to a co-occurring heterozygous duplication that confounded the tool and the other due to incomplete (TG)_mT_n tract detection in the reverse chromatogram. No clinically significant misclassifications were observed. CFTR-TIPS is a free, accurate, and rapid tool for CFTR (TG)_mT_n tract genotyping using cost-effective Sanger sequencing. This tool is suitable both for automated use and as an aid to manual review to enhance accuracy and reduce analysis time.

1. Introduction

Cystic fibrosis (CF) is one of the most common genetic diseases, impacting an estimated 160,000 living patients worldwide [1]. As an autosomal recessive condition, CF is caused by biallelic (homozygous or compound heterozygous) pathogenic variants in the CFTR gene. CFTR encodes a transmembrane chloride transporter, and its dysfunction leads to altered secretions, obstruction, and/or destruction in multiple organs (e.g., lungs, pancreas, and intestine) [2].

Despite significant advancements in therapies for CF patients, the life expectancy of those affected with the most severe form of the disease, also known as “classic CF”, is less than 50 years [3], with respiratory failure being the leading cause of mortality [4]. In addition to “classic CF”, pathogenic variants in the CFTR gene may also cause less severe CFTR-related disorders, such as CFTR-related pancreatitis [5] and congenital bilateral absence of the vas deferens (CBAVD) [6].

In the United States, approximately one in thirty-five individuals carries at least one pathogenic variant in CFTR. These carriers are at risk of having a child with CF if their reproductive partner is also a carrier. Ashkenazi Jewish and European Americans are more likely to be CF carriers, with an estimated frequency of one in twenty-five individuals [7]. Because of the significant carrier rate and the high disease severity, the American College of Obstetricians and Gynecologists (ACOG) recommends that CF carrier screening be offered to all women who are pregnant or considering pregnancy [8].

In clinical laboratories, CFTR sequence analysis is complicated by a region with low sequence complexity in intron 9 of the gene. This region contains a TG dinucleotide repeat followed by a poly-T mononucleotide repeat, hereafter referred to as the (TG)_mT_n tract (Figure 1A). Both (TG)_m and T_n are polymorphic, with most individuals carrying T₇, or “7T” (Figure 1B). While less common, T₅ and T₉ alleles, also known as “5T” and “9T”, have been reported (Figure 1C,D) [9].

The (TG)_mT_n tract is in the splice acceptor region of intron 9 [10] and is responsible for the proper inclusion of exon 10 in the mature mRNA [11]. Exon 10 is required for a functional CFTR protein. The T₉ and T₇ alleles (with any TG repeat size) are clinically benign, as is the (TG)₁₁T₅ allele [12]. On the other hand, T₅, in combination with longer TG repeats, such as (TG)₁₂T₅ and (TG)₁₃T₅, is clinically significant due to substantial exon 10 mis-splicing [9,13]. These alleles are enriched in CBAVD patients [14]. They also act as genetic modifiers that increase the severity and penetrance of the CFTR R117H variant, which, by itself, is a mild and low-penetrance pathogenic variant, in causing classic CF [15]. Moreover, they have been reported to cause classic CF when present in trans with another severe pathogenic variant such as CFTR F508del (a.k.a., ΔF508).

Because only (TG)₁₂T₅ and (TG)₁₃T₅ are considered clinically significant, while (TG)₁₁T₅ is not, the American College of Medical Genetics and Genomics (ACMG) recently recommended that molecular testing laboratories determine and report the TG repeat size whenever T₅ is detected [16,17]. Clinical assays largely use one of two technical approaches to determine the TG repeat size: Sanger sequencing [18] and targeted next-generation sequencing (NGS) [9,19]. Although Sanger sequencing is less expensive and has a faster turnaround time, compound heterozygosity (i.e., individuals with two different (TG)_mT_n alleles) in this low-complexity region complicates review of the Sanger chromatogram (Figure 1C). In our clinically validated workflow, manual interpretation by an experienced technologist is required to resolve the genotypes. In contrast, sequencing reads from NGS can readily resolve the (TG)_mT_n allele genotypes (Figure 1D); however, its higher cost limits widespread application in cost-conscious settings.

Figure 1. Analysis of the CFTR (TG)_mT_n tract using Sanger and next-generation sequencing. (A) Overview of the (TG)_mT_n tract in CFTR intron 9. Dashed underline: the TG dinucleotide repeat. Solid underline: the poly-T repeat. (B,C) Bidirectional Sanger chromatograms for a sample homozygous for the (TG)₁₁T₇ allele (B) or compound heterozygous for the (TG)₁₂T₅ and (TG)₁₁T₉ alleles (C). In (C), the different lengths of (TG)₁₂T₅ (29 bp) and (TG)₁₁T₉ (31 bp) caused overlapping peaks in the Sanger chromatograms, complicating interpretation. (D) NGS reads, visualized using the Integrative Genomic Viewer (IGV) [20], for a sample compound heterozygous for the (TG)₁₁T₇ and (TG)₁₀T₉ alleles. The genotype could be readily resolved using individual reads.

Figure 1. Analysis of the CFTR (TG)_mT_n tract using Sanger and next-generation sequencing. (A) Overview of the (TG)_mT_n tract in CFTR intron 9. Dashed underline: the TG dinucleotide repeat. Solid underline: the poly-T repeat. (B,C) Bidirectional Sanger chromatograms for a sample homozygous for the (TG)₁₁T₇ allele (B) or compound heterozygous for the (TG)₁₂T₅ and (TG)₁₁T₉ alleles (C). In (C), the different lengths of (TG)₁₂T₅ (29 bp) and (TG)₁₁T₉ (31 bp) caused overlapping peaks in the Sanger chromatograms, complicating interpretation. (D) NGS reads, visualized using the Integrative Genomic Viewer (IGV) [20], for a sample compound heterozygous for the (TG)₁₁T₇ and (TG)₁₀T₉ alleles. The genotype could be readily resolved using individual reads.

Here, we present CFTR-TIPS (CFTR Tool for Inferring Poly-T/TG Size) version 1.0, a web-based software tool that automates the inference of the CFTR (TG)_mT_n genotype from bidirectional Sanger chromatograms. This software is compatible with any Sanger chromatogram that contains the (TG)_mT_n tract ± 15 bp. We evaluated CFTR-TIPS using 835 samples previously tested for the (TG)_mT_n tract in a Clinical Laboratory Improvement Amendment (CLIA)-certified, College of American Pathologists (CAP)-accredited clinical laboratory. CFTR-TIPS achieved 99.8% concordance with the clinically validated manual workflow, and there were no clinically significant misclassifications.

CFTR-TIPS enables efficient and accurate inference of the CFTR (TG)_mT_n tract genotype using cost-effective Sanger sequencing. A preview version of CFTR-TIPS can be found at https://qd29.shinyapps.io/cftr-tips/ (accessed on 2 August 2024). Source code is available at https://github.com/qd29/cftr-tips/ (accessed on 2 August 2024) for local implementations.

2. Results

2.1. Architecture of CFTR-TIPS

The input of CFTR-TIPS consists of bidirectional (forward and reverse) Sanger chromatogram (.ab1) files for a given sample in the ABIF format. These files are generated by Applied Biosystems DNA analyzers (Waltham, MA, USA). CFTR-TIPS outputs potential (TG)_mT_n allele combinations (i.e., genotypes) that may match the input chromatograms, ranks them by goodness of fit, and visualizes their expected peak patterns alongside the observed chromatograms. The visualizations assist the user in determining the most likely (TG)_mT_n genotype in the sample. The architecture of CFTR-TIPS is described in detail below and illustrated in Figure 2.

CFTR-TIPS first scans the input chromatograms to locate the (TG)_mT_n tract. In the forward chromatogram (Figure 2A, upper panel), the 5′-most position is anchored using the upstream 15 bp flanking sequence, and the tract ends when the signal from thymine is no longer detected. Similarly, in the reverse chromatogram (Figure 2A, lower panel), the 3′-most position is anchored using the downstream 15 bp flanking sequence, and the tract ends when the signal from adenine is first detected. The end positions may be inaccurate due to potential peak overlaps; nonetheless, this has been factored into CFTR-TIPS. Overall, this approach allows CFTR-TIPS to be broadly compatible with any primer design that sequences the (TG)_mT_n tract ± 15 bp.

CFTR-TIPS then exhaustively generates the expected peak patterns for all possible (TG)_mT_n genotypes in a user-defined search space (Figure 2B). The default space encompasses T₃ to T₁₁ in combination with (TG)₈ to (TG)₁₆, encompassing all known (TG)_mT_n alleles.

CFTR-TIPS next compares the observed and expected peak patterns. Genotypes are eliminated if their expected patterns are incompatible with the observation (e.g., at a given position, a specific nucleotide is expected, but its signal not observed). For the remaining genotypes, CFTR-TIPS calculates their goodness of fit (using a normalized difference score, see Methods) with the observed peak pattern (Figure 2C). Sorted by goodness of fit, the expected patterns for these genotypes are visualized in the graphical user interface (GUI), alongside the observed Sanger chromatograms (Figure 3).

2.2. Graphical User Interface of CFTR-TIPS

The web-based GUI is divided into user input and output sections. Example data from three de-identified samples are also provided in the GUI.

For the user input section (Figure 3A), the forward and reverse Sanger chromatograms, using the .ab1 file extension, are required. The user may optionally re-define the search space of (TG)_mT_n alleles (using the “Minimum # of T”, “Maximum # of T”, “Minimum # of TG”, and “Maximum # of TG” parameters in the “Optional information” section). Given that the default search space encompasses all known (TG)_mT_n alleles, adjustments to these parameters will rarely, if at all, be necessary. The user may also optionally adjust the “Minimum informative Sanger trace signal” parameter. In the chromatograms, positions with signal intensity below this value will be ignored when comparing the observed and expected patterns. We recommend adjusting this parameter based on the overall quality of the user’s Sanger chromatograms.

After the user clicks the “Run Analysis” button, the output section (Figure 3B) visualizes the observed peak pattern of the (TG)_mT_n tract in the uploaded chromatograms, alongside the expected pattern for a given (TG)_mT_n genotype (shown as letters, i.e., T, G, or T/G, below the observed chromatograms, see Figure 3B). The top of the image displays the name of the genotype, its rank among all possible genotypes, its normalized difference score, and additional metadata. By default, the genotype with the best fit (i.e., lowest normalized difference score) is displayed. The user may navigate among all possible genotype using the “Previous” and “Next” buttons at the bottom of the page.

In the image, some positions in the expected peak pattern may be shaded in blue (Figure 3B). At these positions, the expected nucleotide(s) differ among the possible genotypes. Thus, they are highly informative in determining the most likely (TG)_mT_n genotype in a given sample. For example, Figure 3B shows the (TG)₁₂T₅/(TG)₁₁T₉ genotype, ranked first for the uploaded Sanger chromatograms. Except for the 5′-most shaded position in the reverse chromatogram (complicated by overlapping peaks), the observed and expected patterns matched at six other shaded positions. In contrast, Figure 4 shows the (TG)₁₂T₅/(TG)₁₁T₇ genotype for the same uploaded chromatograms, which was ranked fifth. In this image, the observed and expected patterns showed mismatches at five of the seven shaded positions, including four positions (two each in the forward and reverse chromatograms) at which peak(s) from thymine and/or guanine were observed but not expected due to the shorter expected tract length for (TG)₁₂T₅/(TG)₁₁T₇. Based on Figure 3B and Figure 4, it can be concluded that the (TG)₁₂T₅/(TG)₁₁T₉ genotype is the better fit for the observed chromatograms.

2.3. Evaluation of CFTR-TIPS

We assembled a cohort of 835 clinical samples tested at Mayo Clinic between September 2022 and December 2023. These samples underwent Sanger sequencing of the CFTR intron 9–exon 10 junction region. Subsequently, a clinically validated manual workflow was used to determine the (TG)_mT_n genotype. The distribution of genotypes in these samples is shown in

Table 1. Genotype distribution of the cohort used for evaluation of CFTR-TIPS.

Genotype	Total Samples	Concordant (%)	Discordant (%)	Failed (%)
(TG)11T5/(TG)11T5	12	12 (100.0%)	0 (0.0%)	0 (0.0%)
(TG)11T5/(TG)12T5	5	5 (100.0%)	0 (0.0%)	0 (0.0%)
(TG)11T5/T7	495	493 (99.6%)	1 (0.2%) 1	1 (0.2%)
(TG)11T5/T9	76	76 (100.0%)	0 (0.0%)	0 (0.0%)
(TG)12T5/T7	147	146 (99.3%)	0 (0.0%)	1 (0.7%)
(TG)12T5/T9	25	25 (100.0%)	0 (0.0%)	0 (0.0%)
(TG)13T5/T7	19	19 (100.0%)	0 (0.0%)	0 (0.0%)
(TG)13T5/T9	2	2 (100.0%)	0 (0.0%)	0 (0.0%)
T7/T7	21	19 (90.4%)	1 (4.8%) 2	1 (4.8%)
T7/T9	12	12 (100.0%)	0 (0.0%)	0 (0.0%)
T7/T11	2	2 (100.0%)	0 (0.0%)	0 (0.0%)
T9/T9	12	12 (100.0%)	0 (0.0%)	0 (0.0%)
Other 3	7	7 (100.0%)	0 (0.0%)	0 (0.0%)
Total	835	830	2	3

¹ Misclassified by CFTR-TIPS as (TG)₁₁T₅/(TG)₁₀T₆. ² Misclassified by CFTR-TIPS as (TG)₁₀T₃/(TG)₁₁T₇. ³ Contains one sample each of (TG)₁₂T₅/(TG)₁₂T₅, (TG)₁₂T₅/(TG)₁₃T₅, (TG)₁₂T₅/T₆, T₆/T₇, T₆/T₉, T₈/T₉, and T₉/T₁₁.

.

We then analyzed this cohort using CFTR-TIPS. CFTR-TIPS was able to successfully infer the (TG)_mT_n genotype for 832 (99.6%) of the 835 samples. For the three failed samples, CFTR-TIPS encountered errors and was unable to infer the (TG)_mT_n genotype. The error message, in lieu of the peak patterns, was displayed in the output section of the tool (Figure 3B). Two of the failed samples were due to an inability of the tool to detect the (TG)_mT_n tract, and one was because CFTR-TIPS was unable to find a matching (TG)_mT_n genotype.

For the remaining 832 samples, we compared the first-ranked (TG)_mT_n genotype inferred by CFTR-TIPS with that determined by manual review. Reassuringly, the results were concordant for 830 (99.8%) of the 832 samples. One discordant sample (manual: T₇/T₇; CFTR-TIPS: (TG)₁₁T₇/(TG)₁₀T₃, Figure 5A) had a 4 bp duplication in the same Sanger amplicon, which confounded CFTR-TIPS in detecting the (TG)_mT_n tract. The other discordant sample (manual: (TG)₁₁T₅/T₇; CFTR-TIPS: (TG)₁₁T₅/(TG)₁₀T₆, Figure 5B) was caused by the inability of CFTR-TIPS to fully detect the (TG)_mT_n tract in the reverse Sanger chromatogram. Notably, there were no misclassifications of clinically significant results, i.e., (TG)₁₂T₅ or (TG)₁₃T₅.

In addition, while the time burden was not formally assessed, CFTR-TIPS generally took less than 10 s per sample. Overall, using 835 samples with diverse (TG)_mT_n genotypes, we demonstrated that CFTR-TIPS facilitates accurate, rapid, and user-friendly inference of the (TG)_mT_n genotype of the CFTR gene.

3. Discussion

3.1. Our Findings Support the ACMG Recommendations

The ACMG recently recommended reporting the (TG)_m size when T₅ is detected [16,17]. Our findings support these recommendations. In our clinical laboratory, we perform a CFTR genotyping assay for carrier screening and testing of symptomatic individuals. This assay automatically reflexes to the Sanger sequencing-based (TG)_mT_n genotype analysis when a T₅ allele is detected. Thus, the distribution of (TG)_m size of the T₅ alleles within our cohort provides a largely unbiased representation of the population that underwent CFTR variant testing.

Among the 803 T₅ alleles identified in our cohort (out of 1670 alleles tested), only 203 (25.3%) were the clinically significant (TG)₁₂T₅ or (TG)₁₃T₅ allele. This proportion is largely in line with previous estimates [9]. Our finding suggests that the vast majority of T₅ alleles are clinically benign, highlighting the necessity of determining the (TG)_m size for accurate risk stratification of T₅ alleles. Thus, incorporating (TG)_m size analysis into CFTR variant testing workflows not only aligns with the ACMG recommendations but also substantially improves the clinical utility of the assay.

3.2. Limitations of CFTR-TIPS

The two discordant samples reveal limitations of CFTR-TIPS. First, CFTR-TIPS is not suitable for samples in which a heterozygous insertion, duplication, or deletion variant is suspected in the same Sanger amplicon. This can be recognized by overlapping peaks with the 5′ end of the TG tract in the forward chromatogram or with the 3′ end of the poly-T tract in the reverse chromatogram (as shown in Figure 5A). Second, when the GUI indicates that CFTR-TIPS failed to fully detect the (TG)_mT_n tract in the forward and/or reverse chromatogram (as shown in Figure 5B), we recommend discarding the results. In both scenarios, the goodness-of-fit calculations may be confounded, leading to erroneous results. Fortunately, samples that fall within both limitations can be easily recognized and discarded when the CFTR-TIPS GUI is reviewed.

3.3. Suggested Applications and Benefits of CFTR-TIPS

NGS is increasingly used in daily practice for detecting mutations in the CFTR gene, particularly for patients with suspected CF or CFTR-related disorders. As shown in Figure 1D, the (TG)_mT_n genotypes can be readily resolved using NGS-based assays. Nonetheless, the ACOG recommends against full-gene sequencing for routine CF carrier screening [8]. As a result, CF carrier screening tests may be performed using targeted genotyping platforms (e.g., genotyping microarray, MALDI-TOF mass spectrometry, multiplex PCR) instead of NGS. In addition, due to cost considerations, targeted mutation panels may remain the first-line test for suspected affected individuals. As a result, many laboratories, including those in North America and Europe, continue to offer these panels.

In our laboratory, the test volume of the genotyping microarray-based CFTR mutation panel in 2023 was more than ten times that of the NGS-based CFTR full-gene sequencing assay. Since targeted genotyping typically cannot reliably determine the TG repeat size, a supplementary method (e.g., Sanger sequencing of the (TG)_mT_n tract region) is needed to adhere to the ACMG recommendations. Moreover, clinical implementation and validation of NGS require significant capital investments and technical expertise, which may be inaccessible in resource-limited settings. Therefore, we are hopeful that our tool, used in conjunction with Sanger sequencing-based methods, will become and remain an integral part of CFTR molecular diagnostics.

In research and/or resource-limited clinical laboratory settings, CFTR-TIPS may be operated in the fully automated mode. This is because of the very high accuracy of the tool (99.8% in our evaluation) even without manual review. Nonetheless, when possible, a cursory manual review of the CFTR-TIPS GUI is recommended. In our study, the two discordant samples were easily identified during a manual review, resulting in 100% accuracy. In addition, in non-resource-limited settings, CFTR-TIPS may be used to assist review and/or confirm results by laboratory technologists, leading to improved accuracy and reduced reviewer time burden, particularly for rare (TG)_mT_n genotypes.

The development of CFTR-TIPS offers significant benefits for patients. One of the main advantages of CFTR-TIPS is its high accuracy in determining (TG)_mT_n genotypes, even for rare alleles such as T₆, T₈, and T₁₁ and in the fully automated mode. Since only T₅ alleles in combination with longer TG repeats are clinically significant, the high accuracy of CFTR-TIPS is crucial for providing patients with precise variant classification (i.e., pathogenic versus benign) and clinical counseling.

In addition, CFTR-TIPS is designed to integrate seamlessly into routine diagnostic workflows. Its user-friendly GUI allow laboratory technologists to quickly learn and efficiently use the software. This reduces the risk of errors and shortens the turnaround time, compared with manual review of the Sanger chromatograms. Taken together, these features of CFTR-TIPS ensure that patients receive timely and reliable diagnostic information, consequently improving the quality of care and supporting better clinical decision making.

4. Materials and Methods

4.1. Software Development, Implementation, Testing, and Availability

CFTR-TIPS was implemented using the R programming language (https://www.r-project.org/, accessed on 2 August 2024, version 4.3.1). The shiny package (version 1.7.5.1) was used to construct the graphical user interface. Other required R packages included bslib (version 0.5.1), tools (version 4.3.1), and sangerseqR (version 1.38.0) [21], as well as their dependencies. CFTR-TIPS is compatible with Windows, macOS, and Linux operating systems with the RStudio software installed. CFTR-TIPS was primarily tested on a computer with an Intel i5-12600 CPU and 16 GB RAM, running Windows 10 Enterprise, RStudio build 524 (version 2023.06.1), and Google Chrome version 122.0.6261.122.

The creation of CFTR-TIPS incorporated several key considerations to ensure its accuracy, reliability, and compatibility. We developed CFTR-TIPS using the RStudio/Shiny platform due to its free and open-source availability, as well as its broad compatibility across operating systems. Central to CFTR-TIPS is an algorithm designed to accurately identify the CFTR (TG)_mT_n tract by detecting the 15 bp 5′ and 3′ flanking sequences of the tract. The flanking sequence length was carefully chosen to balance sequence uniqueness (i.e., ensuring that they are not found elsewhere in or near the CFTR gene) and maximum compatibility with various PCR primer designs. The sangerseqR package was selected to process the input Sanger chromatograms. Specifically, it converts the input .ab1 files, which are not directly readable by R, into R-compatible data structures. Additionally, sangerseqR performs base/peak calling, which is essential for CFTR-TIPS to accurately detect the (TG)_mT_n tract and perform goodness-of-fit calculations.

The informatics of CFTR-TIPS was also designed to reliably handle the diversity of (TG)_mT_n genotypes in the human population and the technical variations in the quality of input Sanger chromatograms. In particular, the CFTR-TIPS algorithm can recognize individuals heterozygous for alleles with different (TG)_mT_n tract lengths, ensuring that overlapping peaks with the (TG)_mT_n tract flanking regions do not interfere with the goodness-of-fit calculations. Through the “Minimum informative Sanger trace signal” parameter in the GUI, users can optionally adjust CFTR-TIPS to better accommodate the specific signal and noise levels of their Sanger chromatograms by discarding signals below the threshold as noise.

Moreover, CFTR-TIPS was created with user-friendliness as a priority. The software has robust error-handling features to guide users through common issues. For example, it provides clear error messages when the (TG)_mT_n tract is not detected or when no combinations of TG and T repeat sizes in the user-defined search space match the uploaded data. Additionally, the CFTR-TIPS GUI offers instructions and demo files to assist users in troubleshooting. This user-centric approach enhances the overall usability of the software. CFTR-TIPS underwent rigorous testing to ensure its accuracy and compatibility. We evaluated CFTR-TIPS, as presented in this manuscript, using a wide range of Sanger chromatograms, encompassing various (TG)_mT_n genotypes, laboratory instruments, and technologists. Additionally, feedback from initial users was incorporated to improve the GUI design. These comprehensive testing efforts ensure that CFTR-TIPS is a dependable tool for CFTR molecular diagnostics.

A preview of CFTR-TIPS is available at https://qd29.shinyapps.io/cftr-tips/ (accessed on 2 August 2024). Source code is available at https://github.com/qd29/cftr-tips/ (accessed on 2 August 2024). Compared with locally deployed versions (using the source codes), the preview version has several limitations. First, the preview version may be substantially slower and may occasionally encounter errors not attributable to CFTR-TIPS (such as HTTP 504 gateway timeout). Second, the preview version only allows one sample to be analyzed at a time. It is necessary to refresh the CFTR-TIPS webpage before analyzing another sample. Local versions do not have this restriction. Finally, while the preview version does not retain any user data, it is not hosted on a Health Insurance Portability and Accountability Act (HIPAA)-compliant server; thus, we recommend only uploading data from de-identified and/or research samples.

4.2. Goodness-of-Fit Calculation for Possible (TG)_mT_n Genotypes

We quantified goodness of fit using a normalized difference score ($D$), as follows. This score was based on squared Euclidean distance, with lower scores denoting better goodness of fit.

$$D={\displaystyle \frac{\sum _{i=1}^{n_F+n_R}{(O_{Gi}-E_{Gi})}^2+{(O_{Ti}-E_{Ti})}^2}{{2\times (n}_F+n_R)}}+{\displaystyle \frac{100-m_F-m_R}{1\times {10}^5}}$$

(1)

Here, $n_F$ and $n_R$ denote the length of the observed (TG)_mT_n tract in the forward and reverse chromatograms, respectively. $m_F$ and $m_R$ denote the length of the expected tract in the forward and reverse chromatograms, respectively. $O_{Gi}$ and $O_{Ti}$ denote the observed relative signal intensity (i.e., signal intensity of a given nucleotide divided by total signal intensity) of guanine and thymine at position $i$, respectively. $E_{Gi}$ and $E_{Ti}$ denote the expected relative signal intensity of guanine and thymine at position $i$, respectively. If thymine was expected at this position, $E_{Ti}$ was set to 1 and $E_{Gi}$ to 0, and vice versa. When both thymine and guanine were expected, both $E_{Ti}$ and $E_{Gi}$ were set to 0.5.

4.3. Cohort for Evaluation of CFTR-TIPS

To evaluate CFTR-TIPS, we assembled a cohort of 835 clinical samples tested at the CLIA-certified, CAP-accredited Molecular Technologies Laboratory in the Department of Laboratory Medicine and Pathology, Mayo Clinic (Rochester, MN, USA). Most samples were sequenced as a reflex for non-T₇ (particularly T₅) alleles detected by a CFTR genotyping assay.

Bidirectional Sanger sequencing was performed for the CFTR intron 9–exon 10 junction region; subsequently, the (TG)_mT_n genotype was determined by a clinically validated workflow. This workflow was based on manual review of the data by experienced technologists. Due to lack of clinical significance, the manual workflow did not report (TG)_m status for non-T₅ alleles. See below for technical details of PCR and Sanger sequencing.

The genotype distribution of these samples is shown in Table 1. In addition to the T₅, T₇, and T₉ alleles, the T₆, T₈, and T₁₁ alleles were also observed in our cohort.

4.4. PCR and Sanger Sequencing

The PCR primers for the CFTR intron 9–exon 10 junction region were as follows: forward, 5′-CCATGTGCTTTTCAAACTAATTG-3′; reverse, 5′-CCAAAAATACCTTCCAGCACTACA-3′. Universal sequencing adapter sequences were included at the end of both primers. The expected amplicon size was 427 bp. A 10 μL PCR reaction contained 5.72 μL PCR-grade water, 2.00 μL KAPA2G buffer A, 0.20 μL KAPA2G dNTP mix, 0.08 μL KAPA2G enzyme, 1.50 μL primer mix (concentration of each primer: 1.25 μM), and 0.50 μL patient DNA (acceptable concentration: 80–250 ng/μL).

PCR was performed on Applied Biosystems Veriti thermal cyclers with the following program: 3 min at 98 °C, followed by 15 cycles of 30 s at 95 °C, 30 s at 64.5 °C (−0.5 °C per cycle at a 50% ramp rate), and 60 s at 72 °C, followed by 20 cycles of 30 s at 95 °C, 30 s at 58 °C, and 60 s at 72 °C, followed by 10 min at 72 °C, and finally hold at 4°C. After PCR, the amplification products were purified using AMPure XP reagents. Sanger sequencing reactions were performed using universal sequencing primers. Applied Biosystems 3730xl DNA analyzers were used for capillary electrophoresis.

5. Conclusions

In this study, we described and benchmarked CFTR-TIPS, a software tool that infers the CFTR (TG)_mT_n genotype from Sanger chromatograms. When operated fully automatically (i.e., when the first-ranked genotype inferred by CFTR-TIPS was accepted without manual review), it achieved 99.8% concordance (830 out of 832 samples) with the clinically validated manual workflow. In conjunction with cost-effective Sanger sequencing, we are hopeful that CFTR-TIPS will facilitate access to CFTR (TG)_mT_n genotype analysis for more patients, in accordance with the ACMG recommendations.