Fast, axis-agnostic, dynamically summarized storage and retrieval for mass spectrometry data

Abstract

Mass spectrometry, a popular technique for elucidating the molecular contents of experimental samples, creates data sets comprised of millions of three-dimensional (m/z, retention time, intensity) data points that correspond to the types and quantities of analyzed molecules. Open and commercial MS data formats are arranged by retention time, creating latency when accessing data across multiple m/z. Existing MS storage and retrieval methods have been developed to overcome the limitations of retention time-based data formats, but do not provide certain features such as dynamic summarization and storage and retrieval of point meta-data (such as signal cluster membership), precluding efficient viewing applications and certain data-processing approaches. This manuscript describes MzTree, a spatial database designed to provide real-time storage and retrieval of dynamically summarized standard and augmented MS data with fast performance in both m/z and RT directions. Performance is reported on real data with comparisons against related published retrieval systems.

A web-based system for creating, viewing, and editing precursor mass spectrometry ground truth data

“…Users can control how many points are rendered for the given view. In the event the setting is lower than the actual number of points, JS-MS selects a representative subset of points using the weighted striding algorithm described in [ 9 ]. Set view window (see Fig.…”

“…Each query includes a requested limit on the number of points returned, which invokes the server's algorithm for selecting a representative subset of points, allowing for the user to view the characteristics of the data while only seeing a portion of the points in the given (m/z, RT) region. The server implements the MzTree data structure [9], which is a modified R-Tree that organizes the MS1 points in alternating sorting of m/z and RT to provide fast query response whether the data region requested is primarily across m/z, RT, or both.…”

“…JS-MS 2.0 also includes extra controls for the user to modify view parameters such as point threshold, logarithmic height scaling, and label precision. The point threshold is a function implemented in Java that limits the number of points rendered in a given view, selecting a representative subset of points using the weighted striding algorithm [9]. Applying a point threshold allows for faster load time and graph navigation.…”

“…The MzTree data structure is a modified R-Tree [9] that interleaves data partitions sorted by RT and m/z for fast queries in either dimension. The previously published version of the data structure did not include the fields required for annotation (such as isotope trace ID and isotopic envelope ID).…”

Toffee – a highly efficient, lossless file format for DIA-MS

“…Regardless, algorithms are limited to taking slices of the data along the retention time (or spectrum) axis only. In contrast, other attempts such as mzDB [4], mzRTree [5], and mzTree [6], focus on random I/O access through the use of an RTree [7] data structure. This allows the data to be accessed along both the m/z (or chromatogram) axis as well as the retention time axis, at the cost of file size or mass accuracy.…”