Learning score function parameters for improved spectrum identification in tandem mass spectrometry experiments

A cost-sensitive online learning method for peptide identification

BACKGROUND

Post-database search is a key procedure in peptide identification with tandem mass spectrometry (MS/MS) strategies for refining peptide-spectrum matches (PSMs) generated by database search engines. Although many statistical and machine learning-based methods have been developed to improve the accuracy of peptide identification, the challenge remains on large-scale datasets and datasets with a distribution of unbalanced PSMs. A more efficient learning strategy is required for improving the accuracy of peptide identification on challenging datasets. While complex learning models have larger power of classification, they may cause overfitting problems and introduce computational complexity on large-scale datasets. Kernel methods map data from the sample space to high dimensional spaces where data relationships can be simplified for modeling.

RESULTS

In order to tackle the computational challenge of using the kernel-based learning model for practical peptide identification problems, we present an online learning algorithm, OLCS-Ranker, which iteratively feeds only one training sample into the learning model at each round, and, as a result, the memory requirement for computation is significantly reduced. Meanwhile, we propose a cost-sensitive learning model for OLCS-Ranker by using a larger loss of decoy PSMs than that of target PSMs in the loss function.

CONCLUSIONS

The new model can reduce its false discovery rate on datasets with a distribution of unbalanced PSMs. Experimental studies show that OLCS-Ranker outperforms other methods in terms of accuracy and stability, especially on datasets with a distribution of unbalanced PSMs. Furthermore, OLCS-Ranker is 15-85 times faster than CRanker.

Data-independent-acquisition mass spectrometry for identification of targeted-peptide site-specific modifications

We present a novel strategy based on data-independent acquisition coupled to targeted data extraction for the detection and identification of site-specific modifications of targeted peptides in a completely unbiased manner. This method requires prior knowledge of the site of the modification along the peptide backbone from the protein of interest, but not the mass of the modification. The procedure, named multiplex adduct peptide profiling (MAPP), consists of three steps: 1) A fragment-ion tag is extracted from the data, consisting of the b-type and y-type ion series from the N and C-terminus, respectively, up to the amino-acid position that is believed to be modified; 2) MS1 features are matched to the fragment-ion tag in retention-time space, using the isolation window as a pre-filter to enable calculation of the mass of the modification; and 3) modified fragment ions are overlaid with the unmodified fragment ions to verify the mass calculated in step 2. We discuss the development, applications, and limitations of this new method for detection of unknown peptide modifications. We present an application of the method in profiling adducted peptides derived from abundant proteins in biological fluids with the ultimate objective of detecting biomarkers of exposure to reactive species.

COOH-terminal collagen Q (COLQ) mutants causing human deficiency of endplate acetylcholinesterase impair the interaction of ColQ with proteins of the basal lamina

Collagen Q (ColQ) is a key multidomain functional protein of the neuromuscular junction (NMJ), crucial for anchoring acetylcholinesterase (AChE) to the basal lamina (BL) and accumulating AChE at the NMJ. The attachment of AChE to the BL is primarily accomplished by the binding of the ColQ collagen domain to the heparan sulfate proteoglycan perlecan and the COOH-terminus to the muscle-specific receptor tyrosine kinase (MuSK), which in turn plays a fundamental role in the development and maintenance of the NMJ. Yet, the precise mechanism by which ColQ anchors AChE at the NMJ remains unknown. We identified five novel mutations at the COOH-terminus of ColQ in seven patients from five families affected with endplate (EP) AChE deficiency. We found that the mutations do not affect the assembly of ColQ with AChE to form asymmetric forms of AChE or impair the interaction of ColQ with perlecan. By contrast, all mutations impair in varied degree the interaction of ColQ with MuSK as well as basement membrane extract (BME) that have no detectable MuSK. Our data confirm that the interaction of ColQ to perlecan and MuSK is crucial for anchoring AChE to the NMJ. In addition, the identified COOH-terminal mutants not only reduce the interaction of ColQ with MuSK, but also diminish the interaction of ColQ with BME. These findings suggest that the impaired attachment of COOH-terminal mutants causing EP AChE deficiency is in part independent of MuSK, and that the COOH-terminus of ColQ may interact with other proteins at the BL.