Bioinformatics Tools to Identify and Quantify Proteins Using Mass Spectrometry Data

So Young Ryu1    Stanford Genome Technology Center, Biochemistry Department, Stanford University, Stanford, California, USA
1 Corresponding author: email address: clairesr@stanford.edu

Abstract

Proteomics tries to understand biological function of an organism by studying its protein expressions. Mass spectrometry is used in the field of shotgun proteomics, and it generates mass spectra that are used to identify and quantify proteins in biological samples. In this chapter, we discuss the bioinformatics algorithms to analyze mass spectrometry data. After briefly describing how mass spectrometry generates data, we illustrate the bioinformatics algorithms and software for protein identification such as de novo approach and database-searching approach. We also discuss the bioinformatics algorithms and software to quantify proteins and detect the differential proteins using isotope-coded affinity tags and label-free mass spectrometry data.

Keywords

Mass spectrometry

Bioinformatics

Protein identification

Protein quantification

Proteomics

Software

1 Introduction

Global protein expression profiles of biological systems in various phenotypic states can reveal the biological roles of proteins (Tyers & Mann, 2003). Mass spectrometry (MS) coupled with liquid chromatography can help us build such a profile by producing an enormous number of spectra. However, building protein expression profiles from these spectra is not straightforward; rather, it involves various sophisticated bioinformatics algorithms. Without bioinformatics tools, interpreting these spectra would be extremely time consuming and less accurate. In this chapter, we introduce currently available bioinformatics tools to analyze MS data.

This chapter contains three sections. In Section 2, we review how MS generates data for protein identification and quantification. In Section 3, we discuss the bioinformatics algorithms and software to identify peptides and proteins. In Section 4, we illustrate the bioinformatics algorithms and software to quantify peptides and proteins.

2 MS Data

2.1 How MS generates data

The procedure for generating MS data is the following (Fig. 1.1): Proteins in a biological complex mixture are digested by enzymes (i.e., trypsin) into peptides. These peptides are separated based on their hydrophobicity by liquid chromatography and then ionized. Mass-to-charge ratios (m/z) of these ionized peptides are recorded. We call these recorded scans precursor ion spectra or MS1 spectra. Between precursor scans, ionized peptides with relatively high intensities are isolated and fragmented by collision-induced dissociation (CID). We call this isolated ionized peptide a precursor ion, and we call the resulting spectrum tandem mass spectrum or MS/MS spectrum. The instrument produces precursor ion spectra and tandem mass spectra alternatively for a certain period of time, producing thousands of spectra for each experiment run.

Figure 1.1 General MS-based proteomics workflow. The peak with the asterisk was isolated and fragmented to generate a MS/MS spectrum.

Both precursor ion spectra and tandem mass spectra contain several peaks as shown in Fig. 1.1. A peak is characterized by its mass-to-charge ratio (m/z) and intensity value. A precursor ion spectrum contains peaks that represent ionized charge-bearing peptides, while a tandem mass spectrum contains peaks that represent fragment ions of one peptide. (Occasionally, multiple peptides can be fragmented in one tandem mass spectrum. But for simplicity, let us assume that peaks in one tandem mass spectrum are from one peptide.) Precursor ion spectra are mainly used for peptide quantification, while tandem mass spectra are mainly used for peptide identification.

2.2 Tandem MS

Fragment ions in a tandem mass spectrum are predominantly b- and y-ions. (We ignore other ions such as a- and z-ions for simplicity.) If a tandem mass spectrum is produced by the peptide sequence “PEPTIDE,” then b- and y-ions can be expressed as prefixes and suffixes of this peptide sequence, respectively. The sequences of “P,” “PE,” “PEP,” “PEPT,” “PEPTI,” and “PEPTID” can be generated from b-ions for the given peptide. The sequences of “E,” “DE,” “IDE,” “TIDE,” “PTIDE,” and “EPTIDE” can be generated from y-ions. Since each amino acid has a unique mass (except amino acids, “I” and “L”), we can easily identify a peptide sequence from a tandem mass spectrum in the following conditions: (1) m/z values of all b- and y-ions are observed, (2) there are no noise peaks, and (3) charge states of peaks can be determined accurately. However, tandem mass spectra do not contain all possible b- and y-ions, but only some of them. Furthermore, they contain noise peaks, and it is not a straightforward task to determine charge states of peaks, especially for low-resolution tandem mass spectra. In the third section, we address how bioinformaticians solve these challenges in identifying peptides from tandem mass spectra.

2.3 MS quantification techniques

There are several MS quantification techniques. For labeling quantification techniques, there are isotope-coded affinity tags (ICATs), isobatic tags for relative and absolute quantitation (Hardt et al., 2005), stable isotope labeling by amino acids in cell culture (Ong et al., 2002), and the 18O labeling universal standard approach (Qian et al., 2009). Also, there are label-free approaches that use either peptide ion current areas or the number of tandem mass spectra identified for a protein of interest. In this section, we briefly review ICAT labeling and label-free approaches.

The ICAT labeling approach utilizes traditional stable isotope dilution theory in the proteomics and measures protein quantitative changes between two different biological samples by MS (Gygi et al., 1999). Prior to MS analysis, two different protein samples are treated with either light- or heavy-ICAT reagents. Then, two samples are combined, digested by trypsin to produce a peptide mixture, and analyzed by LC–MS–MS. The ratios of peptide intensities labeled “light” and “heavy” in precursor ion spectra provide relative quantification of proteins between two biological samples. Disadvantages of ICATs are that it cannot quantify proteins without cysteine residues and that it can perform only pairwise comparisons of two biological samples. To overcome these limitations, the many variants of ICATs were developed (Patterson & Aebersold, 2003).

In contrast to the ICAT quantification technique, the label-free approach using peptide ion current areas does not label biological samples. In this approach, each individual sample is separately analyzed by LC–MS–MS. Thus, the label-free approach makes it possible to compare more than two samples without limiting the number of samples to be compared. Moreover, it reduces experimental cost by not requiring labeling reagents. However, when analyzing label-free samples, it is important to have very robust performance of instruments in measuring peak intensities and elution time. Producing reproducible data is one of the keys in this approach.

As mentioned earlier, the label-free approach using peptide ion current areas uses peptide intensity values measured in MS1 spectra. In contrast, an alternative label-free approach, called spectral counting, uses MS/MS spectra instead of MS1 spectra (Colinge, Chiappe, Lagache, Moniatte, & Bougueleret, 2005; Liu, Sadygov, & Yates, 2004). This approach sums up all the MS/MS spectra identified for each protein and uses the total number of MS/MS spectra as the relative protein abundance. The MS data for this approach can be prepared in the same way as the label-free approach using peptide ion current areas. The spectral counting approach may not measure protein quantification as accurately as the other quantitation methods, but it does not require highly robust performance as with the label-free approach using peptide ion current areas. Thus, this approach is often used by pilot studies.

In later sections, we introduce the bioinformatics algorithms using MS data produced by the quantitation approaches we discussed in this section.

3 Protein Identification

3.1 Bioinformatics tools for peptide identification

3.1.1 De novo approach

There are various approaches to identify peptides from tandem mass spectra. The de novo approach is not the most popular way, but it is a natural way to interpret tandem mass spectra considering the mechanism that generates fragment ions such as b- and y-ions. It identifies peptide sequences based on masses of b- and y-ions in MS/MS spectra. If the two smallest observed masses of all the possible b-ions of a peptide are 72 and 171 Da, we know that the first two amino acids of this peptide are “AV” using the theoretical masses of 20 amino acids in Table 1.1. In detail, the mass of the first b-ion matches the...