Use of Artificial Neural Networks for the Accurate Prediction of Peptide Liquid Chromatography Elution Times in Proteome Analyses

Abstract

The use of artificial neural networks (ANNs) is described for predicting the reversed-phase liquid chromatography retention times of peptides enzymatically digested from proteome-wide proteins. To enable the accurate comparison of the numerous LC/MS data sets, a genetic algorithm was developed to normalize the peptide retention data into a range (from 0 to 1), improving the peptide elution time reproducibility to approximately 1%. The network developed in this study was based on amino acid residue composition and consists of 20 input nodes, 2 hidden nodes, and 1 output node. A data set of approximately 7000 confidently identified peptides from the microorganism Deinococcus radiodurans was used for the training of the ANN. The ANN was then used to predict the elution times for another set of 5200 peptides tentatively identified by MS/MS from a different microorganism (Shewanella oneidensis). The model was found to predict the elution times of peptides with up to 54 amino acid residues (the longest peptide identified after tryptic digestion of S. oneidensis) with an average accuracy of approximately 3%. This predictive capability was then used to distinguish with high confidence isobar peptides otherwise indistinguishable by accurate mass measurements as well as to uncover peptide misidentifications. Thus, integration of ANN peptide elution time prediction in the proteomic research will increase both the number of protein identifications and their confidence.

An automated high performance capillary liquid chromatography-Fourier transform ion cyclotron resonance mass spectrometer for high-throughput proteomics

“…The modified FTICR instrument was coupled to a very high pressure, high efficiency capillary LC separation and the entire LC-FTICR system was fully automated. Unattended operation of the instrument revealed the exceptional reproducibility (1-5% deviation in elution times for peptides from a bacterial proteome, although it should be noted that these residual variations can be corrected in data analysis [61]), repeatability (10 -20 % deviation in detected abundances for peptides from the same aliquot analyzed weeks apart), and robustness (highthroughput operation for 5 months without significant downtime) of the overall LC-FTICR system. When combined with the modulated-ion-energy gated trapping, the internal calibration of FTICR mass spectra decreased the dispersion of the mass measurement errors for identification of peptides in conjunction with LC separations to high mass accuracies over a dynamic range of 10 3 in each spectrum.…”

“…Second, if the higher abundance peptide was detected in more than 20 consecutive scans, only the isotopic distributions from 20 scans centered at the maximum of the peptide elution profile were then treated in subsequent statistical analysis. Third, tentative peptide "hits" were initially evaluated over a relatively broad mass window of 50 ppm; then observed LC elution times for the peptides were compared to the theoretically predicted values (as recently developed by this laboratory [61]) and an acceptability cut-off of ϩ/Ϫ 5%. This approach proved to be effective in eliminating false positive identifications, and provided unbiased representation of the MMA for the FTICR during LC separations.…”

The Utility of Accurate Mass and LC Elution Time Information in the Analysis of Complex Proteomes

“…The utility of this information increases with the peak capacity of the separations and the reproducibility of peptide elution times [24,25]. Although the absolute LC elution time of a particular peptide can vary from run to run because of temperature and flow rate, among other factors, these changes can largely be corrected after normalization by using an appropriate algorithm to align multiple analyses [26,27].…”

Identification of tryptic peptides from large databases using multiplexed tandem mass spectrometry: simulations and experimental results

“…These pieces of information include the mass to charge ratios (m/z) of the parent species, the connectivity with their fragments (i.e., which fragments come from which parent), the m/z of the fragments, and as we show elsewhere [16], the LC retention time. In multiplexed MS/MS, when multiple ions are dissociated simultaneously, we deliberately lose the explicit connectivity information between parent and fragment ions.…”

“…Moreover, the new statistical approaches that allow correlation of peptides sequence and fragmentation patterns [20] are expected to play a major role in the development of confident identification procedures not only for regular MS/MS, but even more so for its multiplexed counterpart, by taking into account not only the peak's position but their intensity as well. In addition, the use of accurately predicted LC elution times [16] could be an effective way to increase confidence and reduce the false positive rate by further limiting the number of candidates for peptides eluting at a given time. Finally, we strongly believe that with the use of DREAMS [21,22] and related data-dependent methods for smarter selection of peptides for dissociation [1], the benefits of multiplexed MS/MS would be entirely realized.…”