Your Questions Answered – Breath Biopsy VOC Atlas^®

About the VOC Atlas

• Where can I access the Breath Biopsy VOC Atlas®?

We are currently promoting the VOC Atlas to invitees only, to allow us to monitor feedback and prioritize accordingly.  We are rapidly adding new features, but there will remain a distinction between the subset of data that we make publicly available (subject to contracts and consent) and the broader VOC Atlas database (which Owlstone Medical uses for further research purposes). 

• What data does the VOC Atlas contain?

At its core, the VOC Atlas is a database of study findings (compound IDs and feature tables).  It also contains the context needed to interpret those (clinical phenotypes, methods, etc). 

Over time, the majority of feature table data in the VOC Atlas will be validated amounts of high-confidence IDs. We also intend to allow externally generated feature tables to be uploaded in the future, including those using other sampling and analytical techniques. 

• How can I use the VOC Atlas?

You can find some Atlas case studies here. 

• Can I submit my data to the VOC Atlas?

Please contact us if you want to publish your data on Atlas. 

The majority of data currently in the VOC Atlas is from Owlstone Medical-run breath studies. However, we are always interested in speaking to partners with relevant data and may allow users to upload their own datasets in the future. 

• Can I download data from Atlas?

Yes, but please see the VOC Atlas Terms of Use. 

If you are unable to download the data you need, please contact us. 

• What are the IP/data rights associated with the VOC Atlas?

Please see the VOC Atlas Terms of Use.

• Does the VOC Atlas contain personal data?

The VOC Atlas database does contain some personal data, but we take the privacy of volunteers very seriously.  We use measures including pseudonymization and aggregation to protect volunteer privacy; please see our privacy policy and Terms of Use for further details. 

• How have you chosen the list of VOCs to include in Atlas?

The current list of VOCs included in Atlas are VOCs that Owlstone Medical has confidently identified and that are distinguished from background contamination. VOCs are not included if they do not have a sufficient signal-to-background ratio as we cannot confidently say they are breath-borne. VOCs that have been tentatively identified (for example, against the NIST library) are also not included. Work continues to confidently identify as many VOCs on breath as possible and to use different methods to improve the signal to background; This list will expand in the future.

• Does the VOC Atlas contain breath data for other compounds (non-VOCs)?

The majority of data in the VOC Atlas currently comes from breath studies using GC-MS (hence will be volatile organic compounds). However, at its core the VOC Atlas is a database of feature tables (compounds, levels) – there is no explicit restriction to specific compound types or analytical methodology.  Owlstone Medical has a particular interest in hydrogen, methane, and digestive health compounds, so over time we expect the scope of the VOC Atlas to broaden beyond GC-MS VOCs. 

• Does the VOC Atlas contain breath data from other analytical techniques (non-GC-MS)?

The majority of data in the VOC Atlas currently comes from studies using GC-MS.  However, Atlas captures the results of a study using a generic feature table format – essentially a list of the compounds and concentrations seen in each sample.  This allows the VOC Atlas to store and present data from other analytical techniques, provided that the data is of sufficient quality (confident IDs, validated quantification). 

• Does the VOC Atlas contain any non-breath clinical data (blood, feces, etc)?

The majority of data in the VOC Atlas currently comes from breath studies.  However, the Atlas captures the results of a study using a generic feature table format – essentially a list of the compounds and concentrations seen in each sample.  We make some assumptions about the collection methodology in order to present concentrations to users, so in practice, we will only include non-breath data if it directly supports a trial for which we do have breath data (eg, blood headspace measurements for a breath trial). 

• Does the VOC Atlas contain any preclinical data?

The majority of data in the Atlas currently comes from human breath studies.  However, the Atlas captures the results of a study using a generic feature table format – essentially a list of the compounds and concentrations seen in each sample.  We make some assumptions about the collection methodology to present concentrations to users, so in practice, we will only include preclinical data if it directly supports a trial for which we do have breath data (eg, preclinical measurements supporting a breath trial). 

• What is the difference between the VOC Atlas and the NIST/HRAM libraries?

Mass spectral libraries (such as those produced by NIST, or Owlstone Medical’s HRAM) are databases of reference spectra, and are used to tentatively identify peaks within a chromatogram; the complete set of identified peaks for samples is known as a feature table.  In contrast, the Atlas is a database of feature tables, alongside the context needed to interpret those (clinical phenotypes, methods, etc).  Hence the Atlas is instrument-agnostic and can compile learnings from a full range of data sources. 

In order to keep the VOC Atlas reliable, we typically require any compounds reported in feature tables to be confidently identified.  Tentative matches (eg, those applied from the NIST library) are generally not applicable to the Atlas. 

• How will the VOC Atlas account for differences in populations and cohorts?

Future versions of the VOC Atlas will enable users to download breath data. Included in this downloadable data will be pseudonymized information on the subjects themselves: from BMI to disease tags. Users will be able to form their own judgment as to how to cross-compare data.

• Do all the literature associations match Owlstone Medical’s quality criteria (on-breath, confident identification, accurate quantitation)?

No, and in many cases the exact methodology will also not be known to us.  However, we do try to capture relevant quality context from the literature where the authors have included it so that readers can prioritize as needed.  For example, each entry has an associated “ID Match” which specifies whether we believe that identity meets our criteria for confident identification (matched against a standard). 

• Will users be able to upload their own literature references to the VOC Atlas?

The literature associations that you see in the Atlas today have been curated by Owlstone’s biomarker scientists, but we have discussed the possibility of “crowd-sourcing” this data in the future.  We are only likely to do so once we feel we have sufficient measures in place to maintain the quality of those associations.  

• What do you mean by on-breath compounds, and why does it matter?

We consider a compound to be “on-breath” when it has a good signal on breath samples, relative to the background.  Specifically, we assess this using Type 1 (standard deviation), Type 2 (paired t-test), or Type 3 (ROC curve) metrics.  More information on these can be found in our recent publication.  A particular compound may be on-breath in one population/study, but not on-breath in another context (eg, due to changes in breath composition, background distribution, or analytical capability). 

On-breath compounds are particularly important for the VOC Atlas because compounds of interest are often volatile, so sources of background contamination can be significant.  Prioritization of compounds that are substantially above background levels increases our confidence in any findings (alongside the statistical significance and biological relevance). 

• What do you mean by ID confidence, and why does it matter?

This is a measure of our confidence that the “assigned identity of a feature” is the “true compound in the sample”.  We adopt a modified version of the MSI standard proposed by Sumner et al.  For the VOC Atlas, we are most interested in “Validated” identities (Tier 1) and “Putative” identities (Tier 2), since these have been matched against standards.  Note that Atlas does include some literature associations where the ID methodology was unclear or tentative, but these are shown as such. 

ID confidence matters because one of the goals of the VOC Atlas is to improve comparability between datasets; this can only be done if the identities of those compounds are known with confidence. 

• What do you mean by quantitation confidence, and why does it matter?

This is a measure of our confidence that the “assigned concentration of a compound” is the “true concentration of that compound on-breath”.  We are not aware of a standard for this, so have used an analogous system to that for ID confidence.  For the Atlas, we are most interested in quantitation which has been validated at either an analytical level (within calibration curve, etc) or a system level (within performance spec for the entire sample collection and analytical system).  Note that analytical uncertainty estimates often only capture a portion of the true accuracy of breath measurements. 

Quant confidence matters because one of the goals of the Atlas is to improve comparability between datasets; this can only be done if we are comparing on the same scale.  It is often unfeasible or impractical to validate the accuracy of an entire collection and analytical system for a wide range of compounds. 

• Can I submit a correction?

Yes, please contact us if you spot data that you suspect may be incorrect.

• Who can I contact for further information?

Please contact us by email or via links on the VOC Atlas website.