AptaSUITE: A Full-Featured Bioinformatics Framework for the Comprehensive Analysis of Aptamers from HT-SELEX Experiments

摘要

class="head no_bottom_margin" id="sec1title">Main Text>To the editor:The capability of producing and efficiently processing big data has revolutionized virtually every field of science and technology and has enabled the analysis of experimental results at unprecedented resolutions. This trend is also evidenced in the rapid emergence, and subsequent field-wide adoption, of the high-throughput systematic evolution of ligands by exponential enrichment (HT-SELEX) protocol in the study of in vitro selection. HT-SELEX extends the traditional SELEX protocol, aimed at the generation of high-affinity and specificity oligonucleotides known as aptamers against a molecular target of interest, by coupling this technology with HT sequencing. Selection is typically performed in iterations and consists of incubating an initially random pool of sequences with the target, followed by the partitioning, and subsequent removal, of non-affine species while amplifying the remaining pool to form the input to the next cycle as well as the source material for sequencing. The resulting sequencing data, consisting of a representative sample of the pool composition after each round of selection, are consequently analyzed in silico through dedicated algorithmic approaches, whereby aptamers predicted to possess the desired application-specific properties are typically subjected to further in vitro verification and post-processing.Notably, in order to guarantee an efficient and accurate in silico pipeline, these computational methods must be carefully designed to maximize efficiency while scaling well vertically (guaranteeing a proportional reduction in computation time with a growing number of available processing units and memory) as well as with increasing data volume. Ideally, such tools would additionally require a low learning curve for experimentalists, be platform-independent, and provide integrated means of storing and retrieving, interacting with, and visualizing aptamer-related information.Indeed, over the past decade, typical HT-SELEX datasets have grown 200-fold from 10,000–100,000 reads per selection cycle to current sizes of routinely over 20 million reads per round. This development constitutes an emerging barrier for many well-established algorithmic tools devised before the big data revolution but which are still actively used in RNA bioinformatics analysis pipelines. Prominent examples include clustering sequence species into aptamer families related to each other by sequence similarity, , and the elucidation of shared motifs in primary and/or secondary structure evolving throughout the selection. Both analysis tasks are well established for small-scale datasets but rapidly become computationally intractable with increasing data volume when performed with traditional methods., More complex approaches, designed for scalable, HT data processing on multi-core environments, as found in data centers and cloud environments, typically require expert knowledge to set up a sensible pipeline and may depend on numerous, potentially non-portable, third-party software packages, increasing the burden of long term maintainability., , In addition, the resulting processed data are predominantly output in pure text format or stored in relational databases, adding to the stack of challenges in efficiently interpreting the results. Finally, while undoubtedly being of great utility, web-based (and therefore graphical) solutions backed by cloud services such as the Galaxy project, are limited in their flexibility of visualizing and interacting with vast amounts of data as they must adhere to the constraints imposed by current web browsers and technologies.To address these issues, we have developed AptaSuite, a full-featured, open source, and platform-independent software collection for the comprehensive analysis of HT-SELEX experiments. In stark contrast to previous methods, each implementing their individual and frequently rudimentary data workflow, AptaSuite provides a unified and robust framework for managing aptamer-related data and leverages this framework to serve the required data in a standardized manner to any particular algorithm built with the software. In its core, AptaSuite consists of a collection of carefully designed APIs (application programming interfaces) and corresponding reference implementations for facilitating input, output, and manipulation of aptamer data (such as sequences, aptamer counts in individual selection cycles, structure information, and more). On top of this powerful core library, a number of previously published approaches, , , , href="#bib15" rid="bib15" class=" bibr popnode">15 have been implemented from scratch and are now combined into this uniform, easy-to-use framework (see href="/pmc/articles/PMC5992478/figure/fig1/" target="figure" class="fig-table-link figpopup" rid-figpopup="fig1" rid-ob="ob-fig1" co-legend-rid="lgnd_fig1">Figure 1). In particular, the selected methods constitute well-established approaches to analyze HT-SELEX data and are specifically designed to leverage particular properties of aptamers and the SELEX process.href="/pmc/articles/PMC5992478/figure/fig1/" target="figure" rid-figpopup="fig1" rid-ob="ob-fig1">class="inline_block ts_canvas" href="/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=5992478_gr1.jpg" target="tileshopwindow">target="object" href="/pmc/articles/PMC5992478/figure/fig1/?report=objectonly">Open in a separate windowclass="figpopup" href="/pmc/articles/PMC5992478/figure/fig1/" target="figure" rid-figpopup="fig1" rid-ob="ob-fig1">Figure 1The Modularized Architecture of AptaSuiteDiagram depicting the programmatic architecture of AptaSuite. Core libraries for the storage, retrieval, and manipulation of aptamers are accessed through a well-defined API which, in turn, serves data to and accepts data from the algorithms responsible for input, processing, and output of aptamers. Core libraries include efficient solutions for storing primary and secondary structure information regarding the accepted aptamers, a digital representation of the performed selection by storing the experimental setup, as well as information about the performed selection cycles and auxiliary tools, such as secondary structure prediction algorithms, which have been ported to Java to maintain platform independence. The software layer currently features AptaPLEX, a multithreaded demultiplexer for HT-SELEX data; AptaSIM, aimed at realistically simulating the selection dynamics of SELEX experiments; AptaCLUSTER for the efficient determination of aptamer families; AptaMUT, tailored to the identification of mutants with improved binding affinity; and AptaTRACE, an efficient algorithm for sequence-structure motif elucidation utilizing the entirety of the available aptamer pools. Finally, each computational method is accessible either from command line or through the graphical user interface.