Performance evaluation of existing de novo sequencing algorithms

Two methods have been developed for protein identification from tandem mass spectra: database searching and de novo sequencing. De novo sequencing identifies peptide directly from tandem mass spectra. Among many proposed algorithms, we evaluated the performance of the five de novo sequencing algorithms, AUDENS, Lutefisk, NovoHMM, PepNovo, and PEAKS. Our evaluation methods are based on calculation of relative sequence distance (RSD), algorithm sensitivity, and spectrum quality. We found that de novo sequencing algorithms have different performance in analyzing QSTAR and LCQ mass spectrometer data, but in general, perform better in analyzing QSTAR data than LCQ data. For the QSTAR data, the performance order of the five algorithms is PEAKS > Lutefisk, PepNovo > AUDENS, NovoHMM. The performance of PEAKS, Lutefisk, and PepNovo strongly depends on the spectrum quality and increases with an increase of spectrum quality. However, AUDENS and NovoHMM are not sensitive to the spectrum quality. Compared with other four algorithms, PEAKS has the best sensitivity and also has the best performance in the entire range of spectrum quality. For the LCQ data, the performance order is NovoHMM > PepNovo, PEAKS > Lutefisk > AUDENS. NovoHMM has the best sensitivity, and its performance is the best in the entire range of spectrum quality. But the overall performance of NovoHMM is not significantly different from the performance of PEAKS and PepNovo. AUDENS does not give a good performance in analyzing either QSTAR and LCQ data.     

Searching sequence databases via de novo peptide sequencing by tandem mass spectrometry

There are many computer programs that can match tandem mass spectra of peptides to database-derived sequences; however, situations can arise where mass spectral data cannot be correlated with any database sequence. In such cases, sequences can be automatically deduced de novo, without recourse to sequence databases, and the resulting peptide sequences can be used to perform homologous nonexact searches of sequence databases. This article describes details on how to implement both a de novo sequencing program called “Lutefisk,” and a version of FASTA that has been modified to account for sequence ambiguities inherent in tandem mass spectrometry data.     

Sequence similarity-driven proteomics in organisms with unknown genomes by LC-MS/MS and automated de novo sequencing

LC-MS/MS analysis on a linear ion trap LTQ mass spectrometer, combined with data processing, stringent, and sequence-similarity database searching tools, was employed in a layered manner to identify proteins in organisms with unsequenced genomes. Highly specific stringent searches (MASCOT) were applied as a first layer screen to identify either known (i.e. present in a database) proteins, or unknown proteins sharing identical peptides with related database sequences. Once the confidently matched spectra were removed, the remainder was filtered against a nonannotated library of background spectra that cleaned up the dataset from spectra of common protein and chemical contaminants. The rectified spectral dataset was further subjected to rapid batch de novo interpretation by PepNovo software, followed by the MS BLAST sequence-similarity search that used multiple redundant and partially accurate candidate peptide sequences. Importantly, a single dataset was acquired at the uncompromised sensitivity with no need of manual selection of MS/MS spectra for subsequent de novo interpretation. This approach enabled a completely automated identification of novel proteins that were, otherwise, missed by conventional database searches.     

De novo peptide sequencing and identification with precision mass spectrometry

The recent proliferation of novel mass spectrometers such as Fourier transform, QTOF, and OrbiTrap marks a transition into the era of precision mass spectrometry, providing a 2 orders of magnitude boost to the mass resolution, as compared to low-precision ion-trap detectors. We investigate peptide de novo sequencing by precision mass spectrometry and explore some of the differences when compared to analysis of low-precision data. We demonstrate how the dramatically improved performance of de novo sequencing with precision mass spectrometry paves the way for novel approaches to peptide identification that are based on direct sequence lookups, rather than comparisons of spectra to a database. With the direct sequence lookup, it is not only possible to search a database very efficiently, but also to use the database in novel ways, such as searching for products of alternative splicing or products of fusion proteins in cancer. Our de novo sequencing software is available for download at http://peptide.ucsd.edu/.     

Automatic quality assessment of peptide tandem mass spectra

MOTIVATION: A powerful proteomics methodology couples high-performance liquid chromatography (HPLC) with tandem mass spectrometry and database-search software, such as SEQUEST. Such a set-up, however, produces a large number of spectra, many of which are of too poor quality to be useful. Hence a filter that eliminates poor spectra before the database search can significantly improve throughput and robustness. Moreover, spectra judged to be of high quality, but that cannot be identified by database search, are prime candidates for still more computationally intensive methods, such as de novo sequencing or wider database searches including post-translational modifications. RESULTS: We report on two different approaches to assessing spectral quality prior to identification: binary classification, which predicts whether or not SEQUEST will be able to make an identification, and statistical regression, which predicts a more universal quality metric involving the number of b- and y-ion peaks. The best of our binary classifiers can eliminate over 75% of the unidentifiable spectra while losing only 10% of the identifiable spectra. Statistical regression can pick out spectra of modified peptides that can be identified by a de novo program but not by SEQUEST. In a section of independent interest, we discuss intensity normalization of mass spectra.     

Rapid validation of protein identifications with the borderline statistical confidence via de novo sequencing and MS BLAST searches

Protein identifications with the borderline statistical confidence are typically produced by matching a few marginal quality MS/MS spectra to database peptide sequences and represent a significant bottleneck in the reliable and reproducible characterization of proteomes. Here, we present a method for rapid validation of borderline hits that circumvents the need in, often biased, manual inspection of raw MS/MS spectra. The approach takes advantage of the independent interpretation of corresponding MS/MS spectra by PepNovo de novo sequencing software followed by mass spectrometry-driven BLAST (MS BLAST) sequence-similarity database searches that utilize all partially inaccurate, degenerate and redundant candidate peptide sequences. In a case study involving the identification of more than 180 Caenorhabditis elegans proteins by nanoLC-MS/MS analysis on a linear ion trap LTQ mass spectrometer, the approach enabled rapid assignment (confirmation or rejection) of more than 70% of Mascot hits of borderline statistical confidence.     

Peptide sequence tags for fast database search in mass-spectrometry

Filtration techniques in the form of rapid elimination of candidate sequences while retaining the true one are key ingredients of database searches in genomics. Although SEQUEST and Mascot perform a conceptually similar task to the tool BLAST, the key algorithmic idea of BLAST (filtration) was never implemented in these tools. As a result MS/MS protein identification tools are becoming too time-consuming for many applications including search for post-translationally modified peptides. Moreover, matching millions of spectra against all known proteins will soon make these tools too slow in the same way that "genome vs genome" comparisons instantly made BLAST too slow. We describe the development of filters for MS/MS database searches that dramatically reduce the running time and effectively remove the bottlenecks in searching the huge space of protein modifications. Our approach, based on a probability model for determining the accuracy of sequence tags, achieves superior results compared to GutenTag, a popular tag generation algorithm. Our tag generating algorithm along with our de novo sequencing algorithm PepNovo can be accessed via the URL http://peptide.ucsd.edu/.     

Efficiency improvement of peptide identification for an organism without complete genome sequence, using expressed sequence tag database and tandem mass spectral data

We compared peptide identification by database (DB) search methods with de novo sequencing results for proteomics study in an organism without genome sequence information. When the former was done by searching the Expressed Sequence Tag (EST) DB of the sample organism or the NCBI nonredundant (nr) protein DB of green plants using either the MASCOT or SEQUEST software program, it was confirmed that the former is as accurate as the latter. Peptides identified from EST DB were twice as many as those from the nr protein DB, in spite of the fact that the EST DB has less data (26 222 EST) than the NCBI nr protein DB (224 238). This study demonstrates that EST DB with tandem mass spectra can be used reliably for high-throughput proteomics studies in an organism without genome information.     

De novo peptide sequencing via tandem mass spectrometry.

Peptide sequencing via tandem mass spectrometry (MS/MS) is one of the most powerful tools in proteomics for identifying proteins. Because complete genome sequences are accumulating rapidly, the recent trend in interpretation of MS/MS spectra has been database search. However, de novo MS/MS spectral interpretation remains an open problem typically involving manual interpretation by expert mass spectrometrists. We have developed a new algorithm, SHERENGA, for de novo interpretation that automatically learns fragment ion types and intensity thresholds from a collection of test spectra generated from any type of mass spectrometer. The test data are used to construct optimal path scoring in the graph representations of MS/MS spectra. A ranked list of high scoring paths corresponds to potential peptide sequences. SHERENGA is most useful for interpreting sequences of peptides resulting from unknown proteins and for validating the results of database search algorithms in fully automated, high-throughput peptide sequencing.     

Assessing peptide de novo sequencing algorithms performance on large and diverse data sets

De novo peptide sequencing algorithms are often tested on relatively small data sets made of excellent spectra. Since there are always more and more tandem mass spectra available, we have assembled six large, reliable, and diverse (three mass spectrometer types) data sets intended for such tests and we make them accessible via a web server. To exemplify their use we investigate the performance of Lutefisk, PepNovo, and PepNovoTag, three well-established peptide de novo sequencing programs.     

The power and the limitations of cross-species protein identification by mass spectrometry-driven sequence similarity searches

Mass spectrometry-driven BLAST (MS BLAST) is a database search protocol for identifying unknown proteins by sequence similarity to homologous proteins available in a database. MS BLAST utilizes redundant, degenerate, and partially inaccurate peptide sequence data obtained by de novo interpretation of tandem mass spectra and has become a powerful tool in functional proteomic research. Using computational modeling, we evaluated the potential of MS BLAST for proteome-wide identification of unknown proteins. We determined how the success rate of protein identification depends on the full-length sequence identity between the queried protein and its closest homologue in a database. We also estimated phylogenetic distances between organisms under study and related reference organisms with completely sequenced genomes that allow substantial coverage of unknown proteomes.     

Spectral dictionaries: Integrating de novo peptide sequencing with database search of tandem mass spectra

Database search tools identify peptides by matching tandem mass spectra against a protein database. We study an alternative approach when all plausible de novo interpretations of a spectrum (spectral dictionary) are generated and then quickly matched against the database. We present a new MS-Dictionary algorithm for efficiently generating spectral dictionaries and demonstrate that MS-Dictionary can identify spectra that are missed in the database search. We argue that MS-Dictionary enables proteogenomics searches in six-frame translation of genomic sequences that may be prohibitively time-consuming for existing database search approaches. We show that such searches allow one to correct sequencing errors and find programmed frameshifts.     

PEAKS DB: de novo sequencing assisted database search for sensitive and accurate peptide identification

Many software tools have been developed for the automated identification of peptides from tandem mass spectra. The accuracy and sensitivity of the identification software via database search are critical for successful proteomics experiments. A new database search tool, PEAKS DB, has been developed by incorporating the de novo sequencing results into the database search. PEAKS DB achieves significantly improved accuracy and sensitivity over two other commonly used software packages. Additionally, a new result validation method, decoy fusion, has been introduced to solve the issue of overconfidence that exists in the conventional target decoy method for certain types of peptide identification software.     

Informatics for protein identification by mass spectrometry

High throughput protein analysis (i.e., proteomics) first became possible when sensitive peptide mass mapping techniques were developed, thereby allowing for the possibility of identifying and cataloging most 2D gel electrophoresis spots. Shortly thereafter a few groups pioneered the idea of identifying proteins by using peptide tandem mass spectra to search protein sequence databases. Hence, it became possible to identify proteins from very complex mixtures. One drawback to these latter techniques is that it is not entirely straightforward to make matches using tandem mass spectra of peptides that are modified or have sequences that differ slightly from what is present in the sequence database that is being searched. This has been part of the motivation behind automated de novo sequencing programs that attempt to derive a peptide sequence regardless of its presence in a sequence database. The sequence candidates thus generated are then subjected to homology-based database search programs (e.g., BLAST or FASTA). These homology search programs, however, were not developed with mass spectrometry in mind, and it became necessary to make minor modifications such that mass spectrometric ambiguities can be taken into account when comparing query and database sequences. Finally, this review will discuss the important issue of validating protein identifications. All of the search programs will produce a top ranked answer; however, only the credulous are willing to accept them carte blanche.     

The spectral networks paradigm in high throughput mass spectrometry

High-throughput proteomics is made possible by a combination of modern mass spectrometry instruments capable of generating many millions of tandem mass (MS(2)) spectra on a daily basis and the increasingly sophisticated associated software for their automated identification. Despite the growing accumulation of collections of identified spectra and the regular generation of MS(2) data from related peptides, the mainstream approach for peptide identification is still the nearly two decades old approach of matching one MS(2) spectrum at a time against a database of protein sequences. Moreover, database search tools overwhelmingly continue to require that users guess in advance a small set of 4-6 post-translational modifications that may be present in their data in order to avoid incurring substantial false positive and negative rates. The spectral networks paradigm for analysis of MS(2) spectra differs from the mainstream database search paradigm in three fundamental ways. First, spectral networks are based on matching spectra against other spectra instead of against protein sequences. Second, spectral networks find spectra from related peptides even before considering their possible identifications. Third, spectral networks determine consensus identifications from sets of spectra from related peptides instead of separately attempting to identify one spectrum at a time. Even though spectral networks algorithms are still in their infancy, they have already delivered the longest and most accurate de novo sequences to date, revealed a new route for the discovery of unexpected post-translational modifications and highly-modified peptides, enabled automated sequencing of cyclic non-ribosomal peptides with unknown amino acids and are now defining a novel approach for mapping the entire molecular output of biological systems that is suitable for analysis with tandem mass spectrometry. Here we review the current state of spectral networks algorithms and discuss possible future directions for automated interpretation of spectra from any class of molecules.     

A workflow to increase the detection rate of proteins from unsequenced organisms in high-throughput proteomics experiments

We present and evaluate a strategy for the mass spectrometric identification of proteins from organisms for which no genome sequence information is available that incorporates cross-species information from sequenced organisms. The presented method combines spectrum quality scoring, de novo sequencing and error tolerant BLAST searches and is designed to decrease input data complexity. Spectral quality scoring reduces the number of investigated mass spectra without a loss of information. Stringent quality-based selection and the combination of different de novo sequencing methods substantially increase the catalog of significant peptide alignments. The de novo sequences passing a reliability filter are subsequently submitted to error tolerant BLAST searches and MS-BLAST hits are validated by a sampling technique. With the described workflow, we identified up to 20% more groups of homologous proteins in proteome analyses with organisms whose genome is not sequenced than by state-of-the-art database searches in an Arabidopsis thaliana database. We consider the novel data analysis workflow an excellent screening method to identify those proteins that evade detection in proteomics experiments as a result of database constraints.     

ScanRanker: Quality assessment of tandem mass spectra via sequence tagging

In shotgun proteomics, protein identification by tandem mass spectrometry relies on bioinformatics tools. Despite recent improvements in identification algorithms, a significant number of high quality spectra remain unidentified for various reasons. Here we present ScanRanker, an open-source tool that evaluates the quality of tandem mass spectra via sequence tagging with reliable performance in data from different instruments. The superior performance of ScanRanker enables it not only to find unassigned high quality spectra that evade identification through database search but also to select spectra for de novo sequencing and cross-linking analysis. In addition, we demonstrate that the distribution of ScanRanker scores predicts the richness of identifiable spectra among multiple LC-MS/MS runs in an experiment, and ScanRanker scores assist the process of peptide assignment validation to increase confident spectrum identifications. The source code and executable versions of ScanRanker are available from http://fenchurch.mc.vanderbilt.edu.     

Sequence optimization as an alternative to de novo analysis of tandem mass spectrometry data

MOTIVATION: Peptide identification following tandem mass spectrometry (MS/MS) is usually achieved by searching for the best match between the mass spectrum of an unidentified peptide and model spectra generated from peptides in a sequence database. This methodology will be successful only if the peptide under investigation belongs to an available database. Our objective is to develop and test the performance of a heuristic optimization algorithm capable of dealing with some features commonly found in actual MS/MS spectra that tend to stop simpler deterministic solution approaches. RESULTS: We present the implementation of a Genetic Algorithm (GA) in the reconstruction of amino acid sequences using only spectral features, discuss some of the problems associated with this approach and compare its performance to a de novo sequencing method. The GA can potentially overcome some of the most problematic aspects associated with de novo analysis of real MS/MS data such as missing or unclearly defined peaks and may prove to be a valuable tool in the proteomics field. We assess the performance of our algorithm under conditions of perfect spectral information, in situations where key spectral features are missing, and using real MS/MS spectral data.     

Spectral networks: a new approach to de novo discovery of protein sequences and posttranslational modifications

Significant technological advances have accelerated high-throughput proteomics to the automated generation of millions of tandem mass spectra on a daily basis. In such a setup, the desire for greater sequence coverage combines with standard experimental procedures to commonly yield multiple tandem mass spectra from overlapping peptides-typical observations include peptides differing by one or two terminal amino acids and spectra from modified and unmodified variants of the same peptides. In a departure from the traditional spectrum identification algorithms that analyze each tandem mass spectrum in isolation, spectral networks define a new computational approach that instead finds and simultaneously interprets sets of spectra from overlapping peptides. In shotgun protein sequencing, spectral networks capitalize on the redundant sequence information in the aligned spectra to deliver the longest and most accurate de novo sequences ever reported for ion trap data. Also, by combining spectra from multiple modified and unmodified variants of the same peptides, spectral networks are able to bypass the dominant guess/confirm approach to the identification of posttranslational modifications and alternatively discover modifications and highly modified peptides directly from experimental data. Open-source implementations of these algorithms may be downloaded from peptide.ucsd.edu.