HiTRACE-Web: an online tool for robust analysis of high-throughput capillary electrophoresis

Hanjoo Kim, Pablo Cordero, Rhiju Das, Sungroh YoonNucleic Acids Res 2013 · DOI · OpenReview

TL;DR

This paper introduces HiTRACE-Web, a user-friendly online platform that significantly streamlines the analysis of high-throughput capillary electrophoresis (CE) data, crucial for mapping complex DNA and RNA structures. By offering an integrated web interface with automated features and parallel processing, it overcomes the steep learning curve and computational demands of previous command-line tools, making advanced CE analysis more accessible to researchers.

What Problem Does This Paper Address?

High-throughput capillary electrophoresis (CE) is a powerful technique for analyzing nucleic acid structures, particularly when combined with chemical probing methods. These experiments can generate tens of thousands of data points (bands) from a single run, requiring sophisticated computational analysis to extract meaningful structural information. Existing software tools, while functional, often suffer from limitations such as a steep learning curve (requiring command-line proficiency and specific software like MATLAB), limited cross-capillary analysis capabilities, or reliance on proprietary utilities that hinder customization. This informatics bottleneck makes it challenging and time-consuming for researchers to analyze large-scale CE datasets, especially for complex applications like RNA structure mapping, chromatin footprinting, and large-scale RNA design projects.

Methods

HiTRACE-Web builds upon the established HiTRACE (High Throughput Robust Analysis of Capillary Electrophoresis) workflow, which involves several key stages:

  • Preprocessing: This includes selecting signal and reference channels, defining the region of interest (ROI) for analysis (either automatically using edge-detection or manually), and adjusting the baseline.
  • Alignment: To account for slight variations in electrophoretic conditions between capillaries, profiles are aligned. This involves within-batch linear alignment to a reference capillary, followed by between-batch global alignment and piece-wise-linear alignment for fine-tuning local discrepancies.
  • Sequence/Structure & Probe Type Specification: Users provide the nucleic acid sequence and specify the type of chemical modification applied to each capillary. This information helps HiTRACE-Web predict where bands should appear.
  • Band Annotation: This crucial step maps each detected band to a specific nucleotide position in the sequence. HiTRACE-Web introduces an automated band annotation functionality that can assign hundreds of bands in seconds, significantly reducing manual effort. It also provides an interactive interface for manual adjustment and correction.
  • Peak Fitting and Quantification: The software approximates each profile as a sum of Gaussian curves centered at intensity peaks. This allows for the quantification of peak areas, providing the final structural information.

The key innovation is the integration of these functionalities into an interactive, user-friendly web environment, accessible via a web browser. This eliminates the need for local software installation and command-line expertise. Furthermore, HiTRACE-Web leverages powerful multi-core server processors for parallelization, making the online workflow faster than many local implementations.

Datasets

The paper describes the evaluation of HiTRACE-Web using data from an RNA structure mapping study employing the mutate-and-map strategy. This dataset involved a 92-nucleotide RNA sequence treated under six different mapping conditions (including SHAPE, DMS, and reference ladders), resulting in a total of 552 bands. The paper does not introduce a new benchmark dataset but rather uses existing data to validate the performance of HiTRACE-Web against its predecessor, HiTRACE.

Key Insights and Ideas

  • Democratizing Advanced CE Analysis: The most significant insight is the successful translation of a powerful but complex command-line tool (HiTRACE) into an accessible online platform (HiTRACE-Web). This dramatically lowers the barrier to entry for researchers wanting to perform high-throughput CE analysis.
  • Automated Band Annotation is a Game-Changer: The introduction of automated band annotation is a major advancement. Manually annotating hundreds of bands across numerous profiles is extremely laborious. Automating this process, while still allowing for manual refinement, saves substantial time and effort.
  • Web-Based Parallelization for Speed: By utilizing server-side parallel processing, HiTRACE-Web achieves faster analysis times compared to typical local implementations, making it more efficient for large datasets.
  • Robustness and Accuracy Maintained: Despite the shift to a web-based platform and the introduction of automated features, the analysis accuracy and robustness of HiTRACE-Web are shown to be comparable to the original HiTRACE tool.

Key Results

  • High Correlation with HiTRACE: Quantitative results from HiTRACE-Web showed a very high correlation (Pearson's R² values of 0.9996 and 0.9999) with those obtained from the original HiTRACE software for two independent users, indicating no major systematic deviations.
  • Consistent Inter-User Results: The consistency of band quantification between independent analyses using HiTRACE-Web (R² of 0.9993) and HiTRACE (R² of 0.9986) further validates the reliability of both tools.
  • Speed Improvement: While specific speedup numbers are not detailed for all operations, the paper states that HiTRACE-Web runs faster than previous off-line implementations due to multi-core server processors. Processing 52 capillaries with 60 bands each takes "a few minutes" on their server.
  • User-Friendly Interface: The paper highlights the development of an intuitive graphical interface that guides users through the analysis pipeline, making it easier to manage complex CE data.

Limitations and Future Work

  • Post-Processing Tasks: HiTRACE-Web does not handle all post-processing tasks, such as correction for signal decay, background subtraction, and final normalization. These are left to the user, as they can be handled differently depending on the specific application and experimental group.
  • Future Enhancements: The authors mention ongoing development of robust experimental and computational approaches for post-processing chemical mapping data, including secondary structure display and error estimation in automatic sequence assignment. They intend to incorporate these advances into HiTRACE-Web to disseminate them rapidly to the community.

Summary updated

BibTeX
@article{kim2013hitrace,
  author = {Hanjoo Kim and Pablo Cordero and Rhiju Das and Sungroh Yoon},
  title = {{HiTRACE-Web: an online tool for robust analysis of high-throughput capillary electrophoresis}},
  journal = {Nucleic Acids Res.},
  year = {2013},
  doi = {10.1093/nar/gkt501},
}
Figure