Introduction

The Digital Olfaction Hub creates reports for the visualization and statistical analysis of experiments from your data sent to the cloud.

Access & Login

The DOH can be accessed at the following URL: https://hub.aryballe.com/

Log in using the credentials for your OKTA account:

Upon your first connection, the following screen is displayed:

Click on the "Start Trial" button. The page will refresh. 

Now click on the "Go to dashboard" button to access the analysis tool:

Home Page

Overview

• "Go to runs" allows you to browse through your cloud database and start analysis.
• "My data sources" gives cloud database statistics such usage, number of runs and devices, latest used Aryballe Suite version etc
• "My saved sessions" gives access to previous analyses. 
• "My sqlite data storage" allows you to analyze data uploaded from a local database. 

The menu can be accessed from anywhere by extending the panel on the left using the arrow at the bottom.

Account settings

The user setting button at the top right of the page displays the expiration date of your trial ans the version you are currently using. To extend your license, please contact support@aryballe.com. 

Log out

Click on the user account settings button and then on "Logout". All changes are automatically saved (except for locked sessions, see "Session management").

Analyze olfactive data

Browse and select dataset

From the Home page, select a cloud database and click on "Go to runs". 

All your data are displayed in chronological order with newest at the top.

Each column header includes a search bar allowing to look for a specific experiments. Per default the loaded data is from the last 2 weeks, but searches can be broaden by using the temporal filter at the top right of the page :

In the "Flags" columns, three checkmarks give insights into the setup used at the time of acquisition:

More details on the experiment can be diplayed by clicking on the "+" at the left:

Select one or more datasets to include in your analysis.

The number of selected run(s) is displayed at the bottom. You can unselect all at once by clicking on the red cross:

Click on "Continue" to start the analysis of the selected dataset.

Analysis page

Overview

The analysis of the dataset is performed on a single webpage called Session. Various selection tools and corrections can be applied to data and instantaneously displayed on resulting diagrams and charts.

How to use this page ?

1. Using the "Item selection" panel, irrelevent items can be excluded from your session. Selecting an item in this list will display the associated sensograms and signatures.
2. In the "Sensorgram and signature" panel, the analyte zone can be refined if necessary. Humidity and temperature evolution are also shown.
3. Using the top right corner "Dashboard settings" icon, various corrections can be applied on the entire dataset. You can exclude peptide or susbstract a common background VOC from the others.
4. Results and metrics of your experiements are shown as PCA, intensity box plot chart, CQS, confusion and distances matrices.
5. When parameters are optimal for your analysis, you can:
   1. Lock your session to avoid any future changes and share the link to this study, or
   2. Export the session as a CSV or PDF file.

The possibility to select a specific zone of the sensorgram could be useful for NeOse Advance's data to make sure that the olfactive signature is generated from a stable plateau phase (the absorption phase duration can vary according to the samples per example). But more importantly, this tool will be very useful for the Amplifier's data to extract the information from different part of the thermodesorption, especially when we observe a chromatographic effect with a signal with multiple peaks.

Example of chromatographic effect:

Icons of the session top banner

In the middle:

Typical general information on the session is the following:

• Author and owner of the session.
• Description is a free text box to add information on the study by clicking on the pen icon
• Date of creation of the current session
• Last update of the current session

The lock icon freezes and saves all changes of the current session. Nobody except the ower can unlock or delete this session (more details in the "Session management" section).

In the top right corner:

Data selection

There are 3 levels of data:

• datasetincluding all the previously selected runs
  • sample (also called item) uniquely identified by the name
    • measure is a replicate, an odor acquisition of a sample.

Samples and individual measures can be excluded from the session by unchecking them : 

In this example, the following are excluded from the study:

• The first measure of Agrunitrile
• All Octanol measures (no checkmark at the left of the item name)
• Some measure of all samples tested, except of PEA (checkmark next to the sample name).

Sensorgrams and signatures

Sensorgrams can be seen at three levels previously mentionned : for a measure, for the sample, and for a dataset.

Measure level

When clicking on the name of a measure, its sensogram and signature are displayed. A colored dot next to the name indicates which measure is displayed.

Hovering your mouse over the individual curves of the sensorgram provides additional details: (time, intensity value at this point), peptide identification.

Analysis can be customized by refining a zone on the sensorgram. This will directly impact the signature.

Tools for the sensogram can be accessed in the top right corner of the sensorgram chart :

Tools for the sensogram can be accessed in the top left corner of the sensorgram chart :

To use keyboard shortcuts, click on the "Activate keyboard shortcuts" and then:

• Press T to toggle Temperature (Blue dotted line)
• Press H to toggle Humidity (Red dotted line)
• Press the left or rightarrow to move the Analyte zone
• Press Shift and select a zone to move the Baseline

By default, the selected analyte zone of an measure is also applied to all other measures of the same sample. If measures are different from each other, unlock the sensorgram to select a different analyte zone for each measure individually:

Sample level

Click on the name of a sample. Mean intensity sensograms are displayed alongside with signature for all selected measures for this sample.

If some measures have an individual analyte zone, a warning is displayed:

The customization/selection of the analyte zone is possible at the sample level and will be applied to all the measures of this sample, erasing individual item changes.

Dataset level

Click on the dataset: mean intensity sensorgrams are displayed for each sample alongside with signature for all selected measures.

Analyte zone selection at this level applies the same zone to all samples, erasing previous changes.

Odor intensity and Humidity charts

1. Select the intensity source: Sensogram for the odor intensity and Humidity for the relative humidity.

2. Select the intensity type. 

• "Average" is the mean value over the analyte zone 
• "Apex" is the highest (or lowest if negative) value 
• "Area" is the sum of values over the selected analyte zone
• "Full Area" is the sum of values from baseline end to measure end

3. Select the type of chart to display the results:

• Box for a standard statistical box plot
• Bar for a bar chart (1 bar = 1 item)
• Temporal for the evolution in time for each sample (1 point = 1 measure)

PCA and CQS

Principal Component Analysis (PCA) is a statistical tool used for multivariate data analysis. In our case the normalized response of the array of n-biosensors is projected and reduced in a new 2D space where the firsts components explain the majority of the variance between measures.

PCAs allow to quickly visualize separation of samples. Each marker is an acquisition, and each color is a sample. The data points on the two principal components plot that are located close to each other indicate a degree of odor proximity, while distant dots indicate different sample odors. 

A good performance of the device is indicated by markers of a same color tightly clustering together, while markers of different colors should be clearly separated in different clusters. The global CQS, shown on top of the PCA chart, indicates NeOse Advance level of performance, based on the comparison of the variability intra cluster and the distance inter clusters. A CQS equal or superior to 50% indicates a good capability of discrimination.

The x axis presents the projection of the olfactive signature on the first principal component of the PCA (PCA 1), and the percentage of expressed variance (55,1% in this example). Similarly, the y axis presents the projection of each measure on the second principal component of the PCA (PCA 2). 

Matrices

Distance Matrix

A distance matrix is a table that shows the distance between a pair of objects. In our case, it displays the Euclidean distances multiplied by 100 between items in normalized signature space: 

• The off-diagonal values represent the mean distance between two different samples
• The values on the diagonal represent the intra-cluster distance (the mean distance of two points that belong to this cluster) of a sample

Lowest distance is dark blue, biggest distance is dark red.

A good performance of the device is indicated by:

• On the diagonal: distance is low because each cluster is small.
• On other cells : distances are big because clusters are far away from each other meaning that items are differentiated.

Confusion Matrix

The confusion matrix is a specific table layout that allows visualization of the performance of an algorithm trained to classify the different selected samples in DOH. Each row of the matrix represents the sample names that were given during the acquisition of the measures, while each column represents the predicted sample names. The values represent the rate (in %) of measures of each sample that were classified in each group by the algorithm.

• 0% stands for no measure in one group being recognized as its cross-group 
• 100% stands for all  measures of one group being recognized as its cross-group
• Any value in between 0 and 100 indicates a certain "confusion"

A good performance of the device is indicated by:

• On the diagonal: 100% 
• On other cells : 0% 

Dashboard settings

Click on "Dashboard settings" to parameter your study:

The settings panel is diplayed at the right-hand side of the page:

Peptide selection

You can exclude peptides from analysis by clicking on the number of the peptide that you previously identified. 

Unselecting a peptide from the study will:

• exclude the signal from the sensogram for the complete dataset
• recompute the signatures, and thus, the PCA and all metrics of the study.

Substract sensogram profile

You can remove the impact of a background on the signal of all the samples, which can be induced by humidity or a diluent for example (Citroflex or Mineral Oil). The signal measured for this background will be substract to all the other samples of the analysis, this is why it has to be a common background to all the samples.

When selecting a sample:

• The selected sample will not be visible in the study and its average sensogram of selected measures will be susbstracted from all other sensograms for the complete dataset
• The signatures, and thus, the PCA and all metrics of the study will be recomputed.

Confidence level

Define the confidence level for statistical metrics. The greater the confidence level, the bigger the error ellipsis around each cluster on the PCA. 

This parameters impacts the ellipses in the PCA.

Exports

Click on the export icon to generate a PDF or a CSV file.

PDF

Click on "Generate PDF report". 

Customize the title and every text fields as needed. All following modifications are saved for this session. You can come back later and even share it with the session link.

If you want to include Confusion Matrix in your final report, toggle this button at the end of paragraf #1.3:

Customize Sensogram and Humidity Intesity charts:

Toggle this button to include all selected items' sensograms at the end of the PDF report:

Click on "Generate PDF". If Sensograms are included, this may take several minutes.

Once open, download it on your computer to save it:

CSV

Click on "Export CSV data"

CSV for Sensograms

Chose an option:

•  "Multi-file ZIP" for a csv file per measure, all gathered in a compressed folder.
• "Single-file CSV" for all data gathered in a single csv file.

Customize the name if needed.

Click on this download button to download sensograms:

CSV for Signatures

Chose an option:

• "Normalize" for L2 Normalized signatures.
• "Raw" for measured values.

Customize the name if needed.

Click on this download button to download signatures:

Manage your sessions

Overview

From the Home page, go to "My saved sessions".

Your sessions are studies for which you are the creator and only owner. Sessions "shared with you" are someone else's studies you visited before.

Click on the session's name to access to the analysis page: 

To add information on a study, write a description and "Enter" to save it:

Note: the session description can also be modified from the session top button : 

Protecting a session

When a session is locked, no future changes are saved. When a session is unlocked, it can be modified by anybody who has access.

Only the session owner can lock/unlock it. 

Lock and unlock a session by clicking on the lock button either from the session management panel: 

or from the session itself:

When consulting a locked session from another owner, you will see this information : 

Sharing a study

A session can be shared with another user by sending the session's url. Copy link using one of the dedicated buttons:

When a locked session is shared with you, but you want to modify it, duplicate it to create your own copy: 

Delete session(s)

Click on the red bin button to delete a session. 

To delete several sessions at once, select them and click on "Delete selected sessions":

For both way, confirmation is expected. Click "yes":

 When deletion is successful this pop-up is displayed:

Compatibility options

Browse and analyze data from local storage 

To analyze data from a local database called "sqlite", from the Home Page go to "My sqite data storage".

Click here to download a script:

Once downloaded, run it by double-clicking.

Click on More info:

Then click on "Run anyway":

This script will upload data from the file %localappdata%\NeOse Pro\db\database.sqlite. When done this message appears:

Leave this window open.

Go back to DOH page and click on "Retry":

The content of the sqlite DB is diplayed and can be browsed. 

Start an session from a dataset and use the same features than for cloud data. Session is saved and can be found in "Saved sessions" page with "DATABASE.SQLITE" as the source:

Now your session exists, you can close the black window and will not need it anymore to access your session.

end of documment