New in PEAKS Studio
PEAKS X: Introduction Video
This video introduces the major new features provided by the newly released PEAKS X. In PEAKS X, a new peptide feature based approach to de novo sequencing and protein identification has been added, breaking the assumption of one peptide per spectrum. This approach merges peptide feature detection, de novo sequencing, and database search to maximize identification sensitivity and accuracy. In this video, we’ll go over the new features which include:
- MS1 feature-based peptide identification
- Support DIA data with database search (no library required)
- Improved identification of endogenous peptides
- Support ion-mobility LC-MS
Click >>HERE<< to download PEAKS X and request a trial key.
In PEAKS X, a new peptide feature based approach to de novo sequencing and protein identification has been added, breaking the assumption of one peptide per spectrum. This approach merges peptide feature detection, de novo sequencing, and database search to maximize identification sensitivity and accuracy. In this video, we’ll go over the new features which include:
•MS1 feature-based peptide identification
•Support DIA data with database search (no library required)
•Improved identification of endogenous peptides
•Support ion-mobility LC-MS
With this approach, data is loaded into PEAKS, then the LC/MS-MS data is analyzed to identify peptide features. The peptide features are then associated with MS/MS spectra. So, multiple peptide features can be matched to a single MS2 spectrum. This allows PEAKS to attempt to match multiple peptides to a spectrum using: de novo sequencing, PEAKS DB, PEAKS PTM, and SPIDER. This has lead to a significant increase in the number peptide identifications that can be made in a single dataset.
A main reason why this new approach offers a significant improvement is its ability to solve chimeric spectra. This common occurrence in LC-MS/MS data arises when multiple peptide features overlap in the acquisition window. The resultant MS2 spectrum will then contain fragment ions from multiple spectra.
Activating this new method is easy to do. During data refinement set up, click ‘associate features with chimera scans. You also have the ability to restrict peptide feature detection based on a few criteria such as charge and intensity.
Spectra that match multiple peptides can be clearly seen in the peptide table of an identification result. For example, this scan (scan 36982 in demix) has matched four different peptides within a 1% false discovery rate. Click ‘Show LCMS’ to inspect the data associated with this chimeric spectrum. MS/MS scans are represented by blue squares. Peptide features are represented by red circles. Scroll over the peptide features to see the full feature area. Looking at the different peptide features, you can see that the MS/MS scan of interest is in the area of multiple peptide features. So, multiple peptides were acquired in the spectrum.
These results are displayed in an intuitive way in our new feature table present in any identification or de novo sequencing result. Here, results are grouped by detected peptide feature. So, if multiple MS2 scans matched your peptide feature, you can see them in the heatmap view, and scroll through them using the buttons above the annotated spectrum. You can also use the protein, show spec, and show LCMS to jump to detailed views about those specific points. Another great interface improvement can be seen here as well. Notice that the fragment error plot scales directly with the annotated spectrum. So, it’s always easy to visualize the mass error of matched fragment ions. The extracted ion chromatogram of the peptide feature can be seen here as well.
Notice the column for de novo sequences in the table. Here, you can see the best de novo sequencing result for scans that could not be matched by database search. By including this, you now have your de novo sequencing results and database search results in one table. So, it’s easy to see the high quality spectra that could not be matched by database search but have a good de novo sequencing result.
This result table can be filtered using the ‘feature view filters’ in the top left corner of the page. You can limit the page to spectra matching entries in the database using the with identification checkbox, filter by feature area or quality, or choose which identifications to focus on with the radial buttons at the bottom. If you choose ‘only original precursor’, only the primary identification for a spectrum will be displayed. If you choose ‘only from chimera’ it will filter the list to only show identifications that don’t match the precursor of the spectrum, allowing you to focus on chimera identifications.
This approach is especially successful when searching datasets containing endogenous peptides. When testing against previously reported endogenous peptide data, PEAKS was able to expand the total number of identified peptides significantly. Since PEAKS is now feature based, it is able to successfully pick up low abundance peptide features. Further improvements were added to PEAKS specifically for endogenous peptides including re-trained algorithms to consider diverse c-termini expected from endogenous peptide data.
De novo sequencing of DIA Data
Another fantastic opportunity that feature based identification brings is the ability to de novo sequence DIA data. It is even able to analyze DIA data without the use of a spectrum library. By identifying peptide features, and matching them to acquisition windows, PEAKS is able to de novo sequence or identify that peptide feature from a DIA MS/MS scan. Since multiple peptide features are acquired in the acquisition window, PEAKS is able to analyze a single DIA spectrum based on multiple peptide features. Since we’re including de novo sequencing of DIA spectra, we’re removing the bias towards previously reported identifications allowing us to make some novel identifications.
The process is illustrated here. Peptide features are detected and matched to MS2 scans. Then, XICs of the precursor as well as the fragment ions can be drawn. Peptide identification of the different features that match the DIA MS2 spectrum is then performed.
Since a significantly larger number of features can be acquired in a DIA spectrum than a DDA spectrum, some novel concepts are used. These include: retention time prediction of peptides, re-use of fragment ions, and peptide fragment prediction.
We have some DIA specific interface improvements as well. When looking at a DIA identification, the LC/MS heatmap shows the MS/MS scan range and mass acquisition window.
Improved Ion Mobility Support
These improvements also provide increases sensitivity when analyzing ion mobility data. Ion mobility data is also very peptide feature rich. So, using a feature focused de novo sequencing and identification approach greatly improves sensitivity for this type of data.
Thanks for taking the time to check out some of the new features provided in PEAKS X for peptide feature based identification. A free trial of PEAKS is available at our website, bioinfor.com.
Overview of PEAKS Studio 8.5 Features
This video briefly introduced the major features released in PEAKS Studio 8.5. This version provides accurate and sensitive proteomics quantification, while continuously providing state-of-the-art discovery functions, including de novo sequencing, database search, and PTM & sequence variant detection.
PEAKS Studio 8.5 is now available, and we have some great new features to share with you. In this video, I’ll take you through some of the key highlights.
First, I’ll show you the global comparative view. This tool breaks down some of the key identification and quantification details at the sample level, allowing you to quickly observe differences between samples.
We’ve also made some tremendous improvements to PTM profiling. This tool uses the area under the curve of peptide features to determine differences in modification quantities. Now it 8.5, it allows you to observe modified peptide differences across samples in one view. Also, it has been integrated into quantification to allow you to consider the protein quantity’s affect on modification quantity.
A large focus has been put on quantification improvements. The PEAKS 8.5 Q module includes a new approach to SILAC. It allows you to easily integrate advanced methods such as super SILAC. Also, retention time alignment has been integrated to allow us to limit the affect of missing values.
Label free quantification has been improved as well. Peptide based quantification has been integrated to improve accuracy, also allowing you to quantify peptides easily. Outlier removal by clustering has also been integrated to provide a more accurate prediction of protein ratios across a proteome.
In the de novo sequencing and identification tools, machine learning was used to create a fragmentation option that accurately identifies peptides from spectra generated using EThcD fragmentation.
First, let’s take a look at the global comparative view
This button in the protein tab will bring up some extra columns you can add to any identification result. The coverage checkboxes will add the percentage of the total protein sequence that was recovered from each of the samples. So, you can now quickly see which samples are identifying more of the protein sequence. The sample area option provides the sum of all peptide feature areas associated with the protein. Peptide feature area has been found to be related to peptide quantity. So, this gives you a quick idea of the quantity of the protein in each sample. Selecting number of spectra breaks down the number of spectra that matched the protein from each sample. This has also been shown to be useful in determining the difference in protein abundance across samples.
In the peptide tab, similar options are available. You can compare peptide feature areas across samples to quickly get an idea in difference in abundance. Also, you can select number of spectra to see which peptides are found in some samples but missing in others.
Next, let’s talk about the improved PTM Profiling with an example
This experiment compared an untreated antibody to an antibody sample treated with an oxidizing agent. Here, in the combined identification results, you can see positions where oxidization occurred. Clicking the PTM profiling button…brings up a view that will allow you to compare oxidization levels between samples. Drop down menus allow you to select which PTMs to visualize, as well as options to only include fully digested peptides, or all peptides. In the bottom left you can see the total modified peptide feature area in red, and the total unmodified peptide feature area in blue. In the top right there’s an option to select a log10 scale or a percentage scale. The two samples are shown side by side at each position. So, in this case as an example, positions where more oxidation occurs in the treated sample are clearly visualized.
The top left shows a table where each modification site’s profiling details are given. Total feature intensities of modified and unmodified peptides are given for each sample. This table can be exported for further analysis.
In PEAKS 8.5, we’ve even integrated PTM profiling into quantification. Protein abundance can change between samples. So, this gives you the ability to account for these changes in your PTM analysis. Here in the quantification PTM profiling table, there is the extra option to normalize the PTM profiling data according to protein quantity. Further details can be seen in the quantification table as well. If multiple peptides and peptide features were considered, they can be seen here. This is important because often the peptide can be found in multiple charge states. Or, the modification site can occur in different peptides. You can also see an XIC curve is generated from the raw data for quality control of the peptide feature area.
Next, in PEAKS 8.5, we’ve updated our SILAC quantification tool.
The experiment window page has been reorganized to fit closely with a SILAC experiment. Select the set of labels from the dropdown menu that fits your experiment. Then hold shift to select your first set of replicates, and click the top blue button. Do the same with each group. In the bottom left is the condition list, this allows you to rename the condition that is associated with each label. Then, specify the reference condition. With this set up, it’s easy to group together multiple replicates.
Here, in the protein tab, all samples are now compared to the reference sample. So, each protein displays a heatmap (red meaning upregulated, green meaning downregulated by default). So, proteins that are more abundant in the conditions relative to the reference sample are clearly visible. The proteins can be sorted by significance, meaning the likelihood that the change observed is not caused by random chance. A paired t-test is used if one sample is included per condition, ANOVA if multiple replicates are used.
Checking out the intuitive peptide display, it shows all the details you need for inspecting individual results: quantitative details, identification, raw quantitative data, and the extracted ion chromatogram of all available labels are given in one view.
One of the greatest improvements of our SILAC quantification module goes largely unseen. Retention time alignment has been built into SILAC to limit the effect of missing values. For example, looking at the details of this peptide by selecting the all vectors button, and looking at the id count tab, you can see that this vector didn’t have an identification associated with it. With PEAKS 8.5, quantitation could still be performed. The identification was missed because an MS/MS scan was not taken of this peptide feature. However, PEAKS was able to apply the MS/MS identification from another sample here, to allow for quantification, providing a more complete set of data.
We’ve also made great strides in the accuracy of our label free quantification.
One of the main reasons for this is our focus on peptide quantification. In PEAKS 8.5, a peptide quantification table has been added to the results as seen here. If a peptide is present in multiple charge states and at multiple retention times, all the quantification result will be grouped together. To see individual feature vectors, click the ‘all vectors’ button. These peptide quantification results are then rolled up to the protein level, where total peptide information is used for protein quantification.
Changes have also been made to the peptides are selected for protein quantification. Clustering is performed to identify the general trend of the supporting peptides, and then remove outliers. For example, this peptide could have been used for protein quantification. But, in this case, clustering determined that it was an outlier, so it was removed as a candidate. Also, in many cases modified peptides behave differently in the mass spectrometer than unmodified peptides. So, you have the option to exclude modified peptides from protein quantification. With all this in mind, the three peptides with the highest average feature area between the samples are used for protein quantification. The peptides selected for the protein’s quantification are shown in the used column.
I would also like to highlight our addition of an EThcD fragmentation option.
EThcD has been shown to give signal rich spectra with c, z, b, and y ion series within a single spectrum. Using machine learning, our team added a new fragmentation option to identify peptides fragmented by this method accurately.
Some smaller, but quite useful additions have been made to the interface.
For example, with this button, you can switch the spectrum intensity scale from relative intensity to absolute intensity values.
Also, a new PTM localization filter has been added to identification results. This allows you to filter modified peptides based on AScore, or minimum ion intensity localization criteria, making it possible to only allow confidently localized modifications into your final result.
Protein and peptide view display filters have been added as well. This way, you can limit your search results to proteins that contain a certain word in the accession or description. For example, to look for specific types of proteins, or you can search for specific PTMs. The peptide filters have some useful additions as well. T show one, in the quantification peptide filters you can limit the peptides by fold change, or significance.
So, give the new PEAKS 8.5 a try to see how these features can help your research. If you would like to learn more about how PEAKS can bring sensitivity and accuracy to your identifications, download a demo or check out more of our tutorial videos.
PEAKS Q: Labelling Quantification – SILAC
PEAKS Studio 8.5 contains an excellent tool for labelled quantification. This video will highlight some of the benefits of this tool and how to use SILAC in your research.
The PEAKS Q module has an intuitive, accurate, and sensitive approach to SILAC quantification. This video will take you through some of the key features in the algorithm that allow it to accurately quantify differences between samples labels using the stable isotope labeling. It also walks you thought set up and analysis of SILAC labeled data with PEAKS.
Three unique features set PEAKS SILAC a part. Our peptide feature detection takes into account factors unique to SILAC such as: isotope charge distribution, MS1 peak intensity correlation over a retention time range, and expected mass shifts caused by labeling. These details are also used to calculate a quality score, allowing you to make sure only good quality labeled peptides are quantified. Also, using retention time alignment, transfer of ID’s across samples to similar peptide features is possible.
Perhaps most importantly, PEAKS Q for SILAC includes an intuitive interface. It allows you to easily visualize and understand your quantification results. I’ll take you through a quantification project in PEAKS to show how to set up and interpret a SILAC project.
No label, arg 6 and r 10, do they the same charge (can tell by the isotopic distribution). Do they share similar intensity profile across the the retention time range of the peptide feature? Do they fall between expected mass shift caused by labelling? Is that mass shift within an acceptable mass error window? This information in also used to calculate a protein quality score. Using this method, reliable quantification can be performed. For example, here we analyzed a data set with known ratios, and compared to MaxQuant.
The x-axis shows the log2 ratio, the y-axis shows the peptide feature frequency, the green lines show the expected ratio. PEAKS was able to consistently accurately match the expected ratio. The next important improvement to our method was peptide ID transfer to limit missing values. When using multiple replicates, it’s expected that some peptide features will have missing identifications. For example, here replicate __ does not have an MS/MS spectrum associated with either the light or heavy peptide feature. So, it is not possible to infer a peptide from this feature if the replicate is considered by itself.
To deal with this PEAKS uses peptide ID transfer. The first application of this is in peptide feature pairs. For example, here, even though an identified MS/MS scan was not found in the heavy peptide feature, it could be transferred from the light peptide feature. So, quantification within the sample could be performed. Also, ID transfer can be done across samples. Here, replicate one doesn’t have an MS/MS spectrum associated with it. But, replicate 2 does. By aligning the retention time of the peptide feature pairs between the replicates, the ID from replicate 2 is transferred to replicate 1. So, quantification can
still be performed on replicate 1, and a missing value was avoided.
You can tell which run the ID was transferred from using the ‘all vectors’ button in the peptide tab. The matching runs will have the same -10lgP score, m/z, charge, ppm mass error, and retention time.
Now that we’ve seen how PEAKS accurately quantifies peptide feature pairs from SILAC experiments, let’s put it into practice. First let’s look at a typical quantification experiment, comparing treated samples to control samples to find proteins that are showing significant change caused by the treatment. Here, cells were grown in the presence of angiotensin II and compared to a control data set. In three replicates, the treated samples were grown in the presence of the heavy SILAC label. In one replicate, the control was labelled with the heavy SILAC label.
This method can be set up in PEAKS easily. First, create a project by going to the top left corner of the screen and clicking the ‘new project button’. Then, add the data from each replicate to a new sample as shown here. At this point, only the enzyme, instrument, and fragmentation type need to be given as parameters. Then, continue through to data refinement, then identification. If you are using this project creation wizard, you don’t need to add the labels as PTMs. When you set up the quantification method at the next step, the PTMs will be added.
In this case, a new method was added. To add a new method, go to configuration, then labeled Q method, and click the new button in the top right hand corner. Select ‘precursor ion quantification’ from the method type drop down. Then, add rows to add labels. Naming is important here, because all labels with the same name will be added together. In this case, light is unlabelled. The heavy labels are R6 and K8.
Coming back to the project creation wizard, at the quantification step, select ‘precursor ion quantification’ from the left. Then select the label method from the top drop down menu. Set the retention time window, this is the maximum allowable retention time shift between the heavy and light label. 20 seconds is usually enough. Also, it’s best to only allow peptides within 1% false discovery rate to be used for quantification.
Below the parameters, you can organize your experiment. Remembering this experiment’s design, each of the four samples is a replicate of the experiment. So, add them all to the same group by holding shift and selecting all the files. Click the top blue button to add the samples to the right. In the bottom left, you can rename the conditions. Then, specify which condition is associated with the light label, and which condition is associated with the heavy label. The lightest label is always shown on top. This is a useful way to organize the display. Because, in this case, the fourth replicate was reversed, and the light label
is the treated sample, and the heavy is the control. PEAKS will account for this. The last step is to make sure the reference sample is set accurately.
Now, we can take a look at the results. The first thing you will see is a heat-map showing which proteins are up regulated and down regulated. By default, red means up regulated and green means down regulated. This can be adjusted if you prefer in the preferences menu. But, how can we make sure the proteins that we’re reporting have real change between the conditions? We have some filters that can be used to make sure your results are accurate. First, check the feature vector filters. Feature vectors are groups of peptide features used for quantification, in this case it’s the combination of the heavy and light peptide feature for one peptide.
Here, we have some filters based on peptide details that have been found to produce more reliable quantification results. The top is the -10lgP of the identification result, by default it’s set to 1% FDR. The second is quality, which calculates the reproducibility of the peptide feature between the different labels.
It considers features such as: the isotope intensity profile, elution profile, and mass shift. You can set a minimum area, more intense peptide features are more reproducible, generally 1E4 is a good cut off. A charge filter is given because peptides in the range of 2-6 are more reproducible. You can also require that the reference be present, require MS/MS identifications, or require that a certain number of labels are present for each peptide.
Next, you can set the protein filters. The most important is the significance, this is a statistic that determines how likely the change that’s observed is caused by more than random chance. We use -10lgP, so a significance filter of 20 equals a p-value of 0.01 or less. This represents a less than 1% chance that the change that’s observed is random. If one group is used (such as this case) a paired t-test is used to calculate significance. If more than two groups are used, such as a super SILAC experiment, ANOVA is used to calculate significance. You can choose to use FDR instead, which is calculated using an adjusted p-value. Modified form exclusion is also included as an option to increase accuracy. If both an unmodified and modified form of a peptide are found, the total abundance of the peptide is split into two features. So, it’s best to remove it from protein quantification. Once you set these filters, you can be confident that the proteins and peptides shown in the protein and peptide tabs are accurate.
The next step is to check the normalization settings. By default, the total intensity of all quantifiable peptides are used to normalize each sample to a total 1:1 ratio. Use the drop down menu to see the normalization ratios for each sample. Other options are available such as ‘normalize to spike’ referring to a spiked in protein. A list of all quantifiable proteins will be displayed. Select a protein or group of proteins that you know have a certain ratio, and enter the expected ratio in the expected ratio field. Then, apply the changes.
* Use P09601 with peptide ALDLPSSGEGLAFFTFPNIASATK
Here in the protein table, you can see the quantification details of any protein that pass the filters on the summary page. The heat-map displays the ratio of each replicate’s treated sample relative to the control.So, these proteins are consistently up regulated relative to the control. The significance column gives the significance of the quantification result. The -10lgP column gives the significance of the identification result. Select a protein to see its peptide details . Its position in the volcano plot will also be highlighted in blue. Double click on a peptide to see the peptide details.
Here in the peptide tab, you can see all the important details of the peptide’s quantification in one page. The top table gives the quality score, heat map, and ratio of the peptide. The bottom four figures show the best annotated MS/MS spectrum to give an idea of the identification quality. A heat-map of the actual peptide features shows the actual data used for quantification. The pink shape shows the area determined to be the boundary of the feature. This information is used to create the XIC curve showing the area used to quantify the peptide in the bottom right. A slice of the MS full can is shown in the bottom left. This page can be used to make sure the peptides passing the filters are trustworthy.
Once you are sure your quantification results are accurate, go back to the summary page, click the export button, and different html and text options are available for you to export your data for further analysis.
You’re now ready to try SILAC quantification with your data! You can try a free demo of PEAKS at bioinfor.com. If you have any questions, send us an email as firstname.lastname@example.org.