PEAKS – PTM Analysis Webinar
This is the video recording of the webinar hosted on November 16th, 2017.
November 16th’s webinar will focus on the capabilities of PEAKS PTM, New Methods for Post-Translational Modification Analysis by LC-MS/MS. Below are a list of points for discussion:
- A common practice for identification of post-translational modifications (PTMs) with PEAKS database search
• Modification analysis of monoclonal antibodies
• Phosphorylation analysis of proteomes (modified enriched peptides, modified peptides of low abundance)
• Localizing exact modification sites
- Discovery of unspecified or hidden modifications with PEAKS PTM (integrating database searching with de novo sequencing to maximize sensitivity of PTM identification)
- Quantitative analysis of PTMs with PEAKS PTM Profiling
• Modified vs unmodified
• Comparison between multiple samples
- New features of PTM analysis with PEAKS 8.5
• Updated PTM database
• PTM filters on summary page for easy clean-up of confident PTMs
• PTM Profiling in DB, PTM vs Q results (SILAC, LFQ, TMT)
PEAKS 8.5 – Overview Webinar
This is the video recording of the webinar hosted on November 2nd, 2017.
This webinar discusses the unique features of PEAKS Studio, including:
- Protein identification by PEAKS DB
- de novo sequencing of peptides missed by database search
- Single amino acid variant identification with SPIDER
- Finding unspecified PTMs with PEAKS PTM
- Semi-quantitative tools included in PEAKS Studio
PEAKS – Label Free Quantification Webinar
This is the video recording of the webinar hosted on October 18th, 2017.
Welcome to the PEAKS webinar today. Previously I have talked about SILAC quantification with PEAKS Studio 8.5. Today it is my pleasure to show you label-free quantification analysis in 8.5. In this new release, we have made some improvements to label-free quantification in PEAKS Q and added a few new functions to make data analysis easier.
First, the LFQ accuracy and sensitivity are further increased by improving feature detection, selection of unique and the three most intense peptides to represent protein quantity, enabling exclusion of redundant peptides and peptides with both modified and unmodified forms. Secondly, we include one more method of calculating abundance: using intensity. Before PEAKS uses peak area to calculate peptide and protein amount by default. With these improvements, PEAKS LFQ shows better results compared to some other LFQ analysis tools. In addition, sample clustering and correlation analyses can be easily done and visualized. Furthermore, we add spectral counting information in identification results so that users can have a semi-quantitation of identified peptides and proteins in each sample. At the end of the talk, I will do a quick step-by-step LFQ data analysis using PEAKS 8.5.
First, I will start with a brief overview of mass spectrometry-based quantitative proteomics approaches. There are label-based and label-free methods. Solid blue and red squares indicate that samples are labeled. The line shows whether samples are handled separately or combined and treated as one. SILAC (table-isotope labeling with amino acids in cell culture), TMT and iTRAQ are popular labeling-based approaches. PEAKS supports analysis of all these types of data. And in previous webinar, I have mainly focused on SILAC data analysis with PEAKS 8.5. Today, I am going to talk about label-free method. Compared to label-based approaches, label-free quantification workflows are simpler, less time-consuming and inexpensive. However, experimental consistencies in sample handling, liquid chromatograph separation and mass spectrometry performance are particularly critical for accurate and precise quantification.
For label-free quantification, three are two major ways to measure protein abundance:
1) spectral counting, which counts the number of MS2 spectra acquired for a given peptide, then adds up the number for all measured peptides of a protein to indicate its abundance;
2) MS1 signal intensity-based, which extracts ion chromatograms and measures area under the curve or just uses the peak height to reflect abundance.
In PEAKS, we use the XIC method by default. In the new release, we added the maximum intensity method, which is used by MaxQuant software for label-free quantification.
Now, let’s take a look at the LFQ workflow in PEAKS Studio.
There are four major steps: feature detection, retention time alignment and feature matching, abundance calculation and feature ID association. I am going to walk you through these critical steps one by one and explain what new algorithms or functions have been developed and integrated to increase LFQ accuracy and sensitivity in 8.5.
The first step is “feature detection”. Individual peptide ion species can generate a cluster of peaks characterized by isotope patterns and elution profiles, which is called a feature. Accurate and sensitive feature detection can have dramatic impact on the quantification results. With improved feature detection algorithm, PEAKS is able to resolve overlapping features as shown in this example, which is one of the major challenges for feature detection.
The second step is retention time alignment followed by feature matching. This is the pre-requisite to transfer peptide identifications to unsequenced or unidentified peptides, thus increasing the number of peptides that can be used for quantification. The retention time of a peptide feature in different LC-MS runs may vary due to a number of reasons. PEAKS uses a maximum matching algorithm to align retention times, the reference paper info is provided below. After alignment, the same peptide features with close mass to charge values and aligned retention times can be matched from different runs and form, what we call, a feature vector.
Next step is relative abundance calculation.
The relative abundance is calculated by using the peak area under the curve. And all isotopic peaks in an isotope pattern are used for abundance estimation. Their intensities in each scan are summed up to construct the XIC curve.
Here are two XICs of a peptide feature, the green one is from run 1, and pink one from run 2. The peptide abundance ratio between two runs is the ratio of two areas. In PEAKS 8.5, we have introduced a ceiling threshold of ratio. If the reference sample or reference group has peptide or protein abundance of 0, then the ratio is defined as 64, the upper boundary. If the sample is 0 also, then a dash will be given since the ratio cannot be estimated.
Lastly, peptide identifications obtained from the MS2 spectra are associated with the peptide feature. The identification is achieved through a de-novo-sequencing assisted database search procedure in PEAKS.
In data-dependent acquisition mode, not all peptide ions can be selected for fragmentation to obtain confident identifications. For example, in this project, there are four raw MS data analyzed. After alignment, a peptide feature vector forms, indicating the same peptide ion species found in all four samples. MS2 and identifications were obtained from two samples only, as annotated by these solid blue squares and shown in the ID Count column. In the other two samples, no MS2 was obtained. However, the peptide ID was associated with this feature vector, so that the unidentified matched features could also be quantified. In other words, as long as the peptide is identified in one sample, it can be quantified in all samples after alignment.
From here, I am going to show you how PEAKS calculates peptide and protein abundance. I’ll start with peptide abundance calculation first. In this example, this peptide in the peptide table, which is newly added to PEAKS 8.5, had two charge states as shown in the associated feature vector table. The peptide abundance in a sample is the sum of its feature vector abundances. As detailed on the right-hand side, in this project, I analyzed four samples, two disease samples and two wildtype samples. This (middle left) table shows the sample abundance profile of the charge 4 peptide. Below is the sample abundance profile of the charge 3 peptide. The sum of charge 4 and 3 peptide area in disease 1 sample is shown as the peptide abundance in this sample. Since disease 1 and 3 samples are grouped as disease group, and WT 1 and 3 samples are grouped as wildtype group, the feature vector group area is calculated as the average of sample area. Noticing here that the area of 0 is not taken into account. Then, the peptide group area is calculated from the sum of the feature vector group area.
When it comes to estimate protein abundance, using the three most intense peptides is one of the best methods. PEAKS first calculates a protein’s supporting peptide average area across all samples and select the top 3 unique peptides for protein abundance estimation. The summed area of these three peptides is used to represent the protein amount in that sample. And the protein group area is the sum of the peptides’ group area.
In addition, when the sequence of a peptide is a sub-sequence of another peptide in the top 3 list, then the peptide with lower average area will not be used for protein quantification. This is because it is likely that redundant quantity information are provided from these two peptides.
Lastly, post-translational modifications that occur at various levels across samples may affect protein quantity estimation. Therefore we add an option to not use modified peptides. You can find this option on summary page, under protein filter. When this box is checked, if both unmodified and modified forms of a peptide are found, they will not be used for protein quantification. There are two examples as pointed by the blue lines here; If unmodified peptide is not found, and there is only 1 modified form, then this peptide can still be used for protein quantification. However, if more than 1 modified forms are found, then all of them will be excluded.
So far, I have introduced the most basic concepts and functions of LFQ with PEAKS 8.5. Yes, one more point to add, as I talked about at the very beginning, before, PEAKS always uses summed area under the curve of the whole isotope envelop as the abundance calculation method. Now we provide another option of using summed maximum intensity. The abundance calculation method can be configured according to user’s preference and it will be documented in the project description.
By using the improved LFQ in PEAKS, we analyzed a published benchmark dataset and compared our results to MaxQuant LFQ results. The sample was composed of a constant pool of yeast proteins spiked with different concentrations of known human standard proteins. Three replicates were performed for each condition. For this dataset, we used intensity to calculate abundance, the same as what MaxQuant LFQ used. Out of the 50 standard proteins, PEAKS quantified 45 and MaxQuant measured 40. Looking at the 39 common proteins, we calculated the protein abundance difference between two adjacent groups and plotted the ratios on the y axis. The orange bars represent MaxQuant result and blue represent PEAKS. You can see that the median of standard protein group ratios from PEAKS are closer to the expected ratios of 3 between the two groups except for C to B comparison. Furthermore, the protein ratios from PEAKS are more focused around 3 than MaxQuant. This table summarized the statistics, showing that PEAKS gave median ratios closer to the expected value and had smaller variance. Besides better LFQ performance, we integrated more functional tools for easy data analysis.
First, we add sample hierarchical clustering analysis. To turn this function on or off, you can click Preferences button, under General display options, check or uncheck the show sample clusters in quantification heatmap. This is the disease vs wildtype dataset that I used before. After applying ANOVA significance score threshold and protein fold change cut-off, we can easily get a small list of highly differentially expressed proteins between two groups and the clustering tree shows 2 disease samples together and 2 control samples together.
To get a complete list of all quantified proteins, you can set the significance value as 0, and fold change at 1 in protein filters on the summary page.
Furthermore, there is a sample correlation button on the summary page. By clicking this button, you can see a matrix of dot plot showing the correlation between any two samples in a project and their Pearson correlation R values.
Lastly, in 8.5, we also provide spectral counting information in the protein identification results. You can see the identified protein’s sequence coverage, spectral counts, and protein area in each sample in a database search, PTM or SPIDER result node, providing you with some semi-quantitative information. Notice that the protein or peptide area calculated here are not from aligned data, so that retention time and features are not matched across different runs and identifications cannot be transferred to handle missing value issues. More accurate and sensitive label-free quantification is provided in the PEAKS Q module.
PEAKS – SILAC Quantification Webinar
This is the video recording of the webinar hosted on September 6th, 2017.
Welcome to join PEAKS webinar today. Previously my colleague, Dan, has introduced PEAKS Studio 8.5 new features. Today it is my pleasure to show you more details related to SILAC quantification analysis in 8.5.
In this new release, we made some big changes to SILAC quantification. First, we further improved detection of SILAC feature pairs and the calculation of quality scores, thus increasing SILAC quantification accuracy and sensitivity. Secondly, we added SILAC peptide ID transfers between different MS runs, so that after alignment, the SILAC pair can be quantified if it is identified in one MS run. Thirdly, the new design of SILAC Q allows more flexible experimental setting that is SILAC-ratio focused and certain statistical tools are integrated to help data analysis. Three case studies will be shown to demonstrate how PEAKS Studio can help you with your SILAC data analysis. The first example is to identify proteins that are consistently changed between labeling channels, for example, heavy and light labeling states, across replicates, which we call single-group SILAC analysis in PEAKS. The second case aims to find whether SILAC ratios change between multiple groups, let’s say, you have drug treated cells and control cells, you want to know at different time points after drug treatment, how the protein levels change between treated vs. control groups, or if you study protein turnover rate over time, or super-SILAC data analysis, etc. The third one is a modification profiling study by using SILAC method. Lastly, I will do a quick step-by-step SILAC data analysis from creating new project to set-up SILAC Q in Studio 8.5.
First, I will start with some SILAC data analysis basics with PEAKS Q.
Here is a SILAC 3-plex data. Proteins were labeled with Arginine light, arginine 6 and arginine 10 and then combined. These three individual features are associated as SILAC pairs based on: first, they are of same charge, which we can tell from the isotopic distribution; secondly, whether these feature peaks share similar MS1 peak intensity profile over retention time; lastly, if they fall within expected mass shifts due to labeling and are within certain mass errors.
Therefore, the higher quality score, the more quantifiable peptide is. Below I am going to show you three SILAC feature pairs assigned with high, medium and low quality scores. The first feature pair has a quality score of 95, which is a good example. You can tell from this figures that this pair has relatively high intensities, similar elution profile. The quality score calculation also considers mass difference, isotopic distribution, and retention time difference. The second one has a quality score of 75, intensities are not very high and XIC shape is not so good. The last one has a quality score of 46, the XIC looks even worse. So then you may wonder if there is a quality score threshold that you can set to filter out bad SILAC feature pairs like this thus not using them for quantification? Yes, we do have this filter on the quantification result summary page; here, you can click the Edit button to open the feature vector filters, then you can set the quality score cut-off here. The range of the score is from 45 to 100. And we recommend to set the cut-off at 50.
By using the improved SILAC feature pair detection, we analyzed a SILAC dataset that had heavy and light peptides mixed at known ratios, marked by the dash green line and compared PEAKS result to MaxQuant result. As shown in this figure, PEAKS gave SILAC ratios that were closer to the expected log2 ratios at 2 than MaxQuant.
PEAKS and MaxQuant are centered on approximately the same ratios and have similar peak widths. Furthermore, PEAKS showed even closer ratios to the expected log2 values at 2.
Next, I am going to talk about peptide ID transfers to handle missing value issues.
Like I introduced before, PEAKS Q detects and associates 2 or 3-plex SILAC feature pairs that have the same charge, similar MS1 peak intensity profile over retention time, expected mass shifts caused by labeling and fall within certain mass errors. If an identification is obtained from one of the labeling states, then the whole SILAC pair can be quantified and used for peptide and protein ratio calculations. For example, the light form of peptide GLGDCLVK was fragmented and identified in sample R2. Although no MS2 spectrum was obtained from the K8 labeled heavy counterpart, the SILAC pair could still be quantified in PEAKS Q as highlighted in the feature vectors table. “Id Count” indicates the number of MS2 identified for each SILAC pair. In addition to this, in 8.5, Peptide IDs can be transferred from a another MS run after alignment. For instance, in sample R_1, there was no MS2 on either heavy or light form of this SILAC pair with a good quality score. To retrieve identification information, retention times of different LC-MS runs are first aligned, then MS2 ID can be matched from another run (for this case, sample R2 run) by aligning SILAC features within tight mass ranges and retention times, thus allowing quantification of SILAC pairs without any ID. And you can tell from which run we transferred the ID by looking at the columns like m/z, z, ppt, RT, -10logP. since these contents will be transferred from the matching run.
From here, I am going to show you three examples of how to use PEAKS Q to analyze complex SILAC data using some published datasets. The first study is to identify proteins that showed consistent expression changes between labeling channels across replicates. The second study is to see how SILAC ratios differ between multiple groups, like a time-series profiling or super-SILAC data analysis. The third study focused on PTM profiling analysis. Before we go to see these specific cases, I would like to first emphasize that in 8.5, SILAC Q result is ratio-focused, and I am going to explain this in the next 2 slides.
First, we use median of peptide ratios to calculate protein ratios. Let’s look at this simple case first, where there is only 1 sample (called R1) in group 1. For this protein, there are 4 supporting peptides. When you mouse-over the heatmap, you can see the peptide area in control and treated samples respectively, this star indicates that the control sample is specified as the reference, so that the ratio is calculated by using 1.2E9 divided by 3.2E8. The peptide ratio is also shown in the group 1 ratio column. When calculating this protein’s SILAC ratio, we are taking the median of supporting peptide SILAC ratios, so that you get 3.76.
When you have multiple samples in one group, like here, R1, R2, R21, in in group 1, this protein’s SILAC ratio in group 1 is calculated from the median of the SILAC ratios in these replicates and shown in the group 1 protein ratio column. Ok. Now we are ready to go to see the three case studies.
The first cast is a typical, we call, single group SILAC data analysis. For this case, usually, you have control and treated samples labeled by SILAC light and heavy medium respectively, and you do 2 or more repeats in order to identify proteins that show different abundance levels between control and treated samples, and also this fold change is quite consistent across technical replicates. For this scenario, we used data from this published paper, the author studied angiotensin 2 (Ang2)-regulated proteomes in kidney cells.
Human primary renal cells were either cultured in regular medium or in 13C6-arginine (R6) and 13C615N2-lysine (K8) SILAC medium. After 6 passages, AngII was added to one cell population and proteins of the treated and control cells were extracted and mixed at 1:1 protein ratio. In total, four replicates were performed, in one of which cells grown in regular medium were treated with AngII. According to this design, here is how I set up the experimental settings in PEAKS Studio 8.5. Two conditions are specified as control and treated. In Sample 1 to 3, light signal is from the control, and heavy from the treated. Whereas in sample 4, the 2 conditions swapped. And control condition is specified as the reference condition so that SILAC ratios of the treated relative to control are calculated for peptides and proteins.
Since the same amount of light and heavy proteins were mixed, auto normalization was first performed in PEAKS Q so that the total light and heavy intensities in each sample were equivalent. The adjusted normalization factor is displayed for each sample in the normalization setting window. Normalized intensities are calculated from raw intensities divided by the normalization factor.
To identify proteins that have significant changes between treated and control conditions across samples in 1 group, paired-t-test is integrated for the significance test. Here in the summary page, it shows that, for 1 group SILAC data, you are using paired-t-test method. The significance score is calculated as -10 times log of the paired t-test p value. In protein filters, you can set the significance score higher than 13, which equals a t-test p value smaller than 0.05. Furthermore, you can require a fold change or the group ratio between 2 conditions higher than a number, for example 1.5 here, to get a smaller list of proteins that have significant differential expressions. We hope that this new design can help you analyze your SILAC data more easily.
The second scenario is called multi-group SILAC data analysis. For this case, usually, you are trying to assess whether protein ratios between different labeling states change across multiple groups. In this published work, the major goal was to identify signature proteins that could differentiate 2 major subtypes, ADC and SCC, of lung cancers.
Four lung cancer patient tumors were injected into 4 mice and the light proteins from the xenografts were analyzed by using the so-called super-SILAC approach. Since patient tumors cannot be labeled, a few lung cancer cell lines were labeled by R10K8 SILAC medium and heavy proteins were combined as the super-SILAC standard, which was spiked into each xenograft sample at 1:1 protein ratio, then analyzed by LC-MS/MS. 3 replicates were performed for each tumor. In PEAKS setting, I put 2 groups, one is the ADC group that includes 3 ADC1 samples and 3 ADC2 samples. The other group is SCC that includes 3 SCC1 tumors and 3 SCC2 tumors. The light is from tumor and the heavy from the standard. By having this super-SILAC standard as a constant reference, proteins from tumors are actually normalized by the standard and what we try to get is the proteins having ratios different between ADC and SCC groups.
For the multi-group SILAC data, PEAKS provides Welch ANOVA test method by default. This method tests whether the protein ratios in one group is significantly different from other groups. You can use either the ANOVA test p values or use a BH-adjusted p value cut-off. Here the fold change means the fold change between 2 group ratios (mouse pointing).
The last case study focuses on PTM profiling analysis between samples or across groups. The demo data aimed to map kinase activation state of an oncogenic tyrosine kinase, Bcr-Abl, in leukemia cells. This fusion protein has 2 major isoforms: p210 and p190. p190 is ~25% shorter than p210 due to a lack of a DH–PH domain unit; otherwise they have an identical sequence and domain organization.
The author put human P210, P190 into a murine pro B-cell line, BaF3, and grew control, p210 and p190 cell lines in SILAC light, medium and heavy medium respectively. They combined three populations of cells, digested proteins to peptides and enriched for phosphopeptides by using p-Tyr antibody and TiO2 beads and analyzed by using LC-MS/MS. Two experiments were performed, each with two technical replicates. In the second experiment, labeling conditions for p210 and p190 swapped. This is how I set up the PEAKS Q accordingly. Three conditions are specified here. Each group includes two repeats. And in Exp2, medium channel is p190, heavy channel is p210.
In the protein coverage view, you can click this “PTM Profile” button to view quantitative PTM results in a new window. For each PTM site, PEAKS calculates the summed MS1 peak area of peptides having this confident PTM site. You can use either A-score cut-off or a pair of supporting fragment ions with minimal relative intensity of the number you set here to filter for confident modified peptides for PTM quantification. Now let’s take a look at what is displayed in the PTM profile pop-up window.
In the top-left table, it lists all confident PTMs, based on either all peptides or fully digested peptides only which can be specified here. For each PTM site, the SILAC ratios of modified peptide area in each labeling condition relative to the reference are calculated and shown. By selecting each row in the table, you can see modified peptide details, and its associated SILAC features of different charge states and in different samples. The XICs of each selected feature are displayed in this figure. You can export the result into a .csv file to see more detailed information.
PEAKS Q: Label Free Quantification
This video jumps straight ahead to asking the question of “How do I analyze a set of data using PEAKS?”. Basics of the software are addressed quickly as to maximize attention towards result explanation and value. We even explain how to use PEAKS label-free quantification to identify peptide features without database peptide identification.
I will now demonstrate how to use PEAKS label-free quantification tool with the real dataset. The first step is to build a PEAKS project for the experiments.
Give a name to the project and specify a location to save the project.
Click the button of “Add data” to add the data.
Navigate to the LC-MS/MS data location, and select the data to be analyzed.
Select a file or files: use the add file button, and give a name for the file to show in the result.
Set enzyme, instrument, fragmentation and continue until all files are added.
After this, the data selection part is finished.
“correct precursor” is checked with “mass only” option for Orbitrap instruments.
Precursor mass error: 20 ppm, Fragment mass error: 0.5 Da
Modification: Fixed: Caramidomethylation
Var: Oxidation on M & Deamidation
Database: Swiss-Prot; Taxa: mouse
Complete identification (DB + PTM + SPIDER)
Label-free quantification parameters:
Estimated peptide precursor m/z shift between samples: 20 ppm
Estimated retention shift of a peptide precursor between samples: 5 minutes
The “Sample Groups”: each sample ran in triplets. Those replicates are grouped.
We have now finished setting the parameters and are ready to import data then perform the analysis. Click “Finish”.
Analysis & Results
After completing analysis, identification and quantification results were present.
Click node 11 and the details of quantification result will be present.
Protein Heat map shows protein abundance profiles among all samples.
These proteins are up-regulated in diseased samples. And their abundance is consistent among replicates.
These proteins are down-regulated in diseased samples. Also their abundance is consistent among replicates.
This is the volcano plot of quantified proteins, which plot significance versus fold-change for proteins.
Three groups: the Majority of proteins are 1. background proteins, 2. Up-regulated proteins which have sufficient up fold change and large significance, 3. And down-regulated proteins which have sufficient down fold change and large significance.
Here are the histograms of feature retention time shifts and precursor m/z shifts, respectively. The red one identifies the shift before retention time alignment and blue one is after alignment.
On the top, there are 2 sets of filters: peptide feature level and protein level.
Significance of peptide feature, fold-change of feature, quality of feature, intensity of feature, etc are used to filter out the peptide features. Protein significance, protein fold-change, and a number of unique peptides in a protein are used to filter out the proteins.
In the protein tab, a list of quantified proteins were displayed, sorted by significance. For each protein, its abundance profile in all samples and groups, etc are displayed. The coverage map shows quantified peptides supporting to this protein.
The details of each peptide feature is listed in the “Feature” tab. The top three peptides with the highest intensities are used for protein ratio calculation. PEAKS also provides peptide feature quantification. The Feature tab shows the list of quantified features.
For each peptide feature, the XICs in all samples were shown, and the integration areas of its XICs are listed in the table.
The characterization of the feature in all samples are shown in the “Sample Feature” tab. The red line shows the boundary of the feature, the blue square is MS/MS spectrum providing peptide identification.
It also provides 3-D view. Press “Ctrl” button, using wheel of mouse to adjust.
No ID? No Problem!
PEAKS can quantify peptide features without the pre-requisite of peptide identification, by un-checking “With peptide ID”.
Here is one example: there is no database peptide hit for this feature. PEAKS DB provides more answers by integrating with de novo sequencing. Let’s look at the de novo sequencing result from PEAKS database search.
Open SPIDER node 10. Select “LC-MS” tab, go file of SMA 1.
LC-MS heat map: PEAKS associates all identification and quantification results with peptide feature in the heat map.
Blue squares indicates the MS/MS spectra.
Solid blue squares are the MS/MS spectrum with a confident database peptide hit.
Solid orange squares show the MS/MS spectrum without a confident database peptide hit, but with a confident de novo peptide sequence.
The red circles are the detected features. Solid ones are the confident features which satisfy the filters. Our purpose is to see if the feature (514.7 17.03) has a de novo sequence. Note that the heat map can zoom. In this case, we use the search function. PEAKS will automatically zoom to corresponding window in the heat map.
PEAKS DB provides more answers by integrating with de novo sequencing. Let’s look at the de novo sequencing result from PEAKS database search. Here is one example where there is no database peptide hit for a feature. That said, there is a de novo sequence: ALNAAGASEPK.
When we BLAST it, and we find that it is a peptide from Myosin-binding protein C, which is a myosin-associated protein found in the cross-bridge-bearing zone of A bands in striated muscle. Myosins comprises a family of ATP-dependent motor proteins and are best known for their role in muscle contraction. With this finding, we get a new biomarker candidate by using de novo only peptides.
You will also note that PEAKS supports the Html format for easy web viewing and text format for sharing and down stream analysis.