Continuous Identification
Continuous Identification¶
GNPS' continuous identification aims to bring datasets to life. Traditionally, after manuscript publication, mass spectrometry datasets remain static and are never heard from again, silo'd in data cemeteries. GNPS' approach is different, we aim to
- Automatically make new discoveries within each dataset
- Aid future analyses by putting new data in the context of all previously existing datasets
GNPS achieves the first goal by periodically reanalyzing each and every public dataset. With ever expanding reference spectral libraries, GNPS makes new identifications within datasets in an automatic fashion, recording this growth in identifications, and reports it back to the community.
GNPS achieves the second goal by enabling users to "BLAST" MS/MS spectra against the entire GNPS set of datasets (see MASST Search).
Exploring Current Continuous Identifications¶
On each dataset page (See datasets) the list of all continuous identifcations is listed under the "Continuous Identification" section. This list of all analyses shows the growth of identifications over time as well as how the identifications changed over time. For example, the 3D molecular skin cartography dataset (See MSV000078556) grew from 94 (April 9, 2014) identifications in the initial continuous identification to 860 (June 1, 2018).
Dataset Subscriptions¶
Whenever new identifications are made in a dataset, subscribers are automatically notified via email. To be subscribed to a dataset of interest, click the subscribe button on the dataset page:
Continuous Identification Ratings¶
Help us improve Continuous Identification by rating the matches returned on the "All Identifications Page". For each identification made, please rate each match according to the following rubric
| Rating | Description |
|---|---|
| 4 stars | correct match as context is right (i.e., molecule is known/expected to be in the sample) |
| 3 stars | compound class match – at least part of the structure makes sense to match |
| 2 stars | cannot tell – might be correct from the spectrum match and context but there is not enough information to tell |
| 1 star | incorrect: molecule does not make sense in this context |
Users will be able to both add their own rating, add a comment for their given rating, as well as view the average rating of the identification.
Continuous ID Subscription Email¶
This example continuous identification email informs subscribers that the dataset of interest has more IDs in the most recent round of continuous identifications. It will first list the title of the dataset, then the changes in identification counts, and finally direct links to explore the data. Users can go directly to the search results and view the new, different, and deleted identifications as well as go straight to the dataset page itself.
Download Living Data¶
This dataset: MSV000084314 includes molecular networks, clustering, and library identifications.
Organization of the FTP Repository¶
They are organized respectively in 4 folders:
- CLUSTERSUMMARY - Represents the clusters for each dataset and the associated metadata (e.g. mz, rt, and number of MS2 spectra)
- CLUSTERINFO - Represents the clustering enabling tracing back from the clusters to the original files and scans they came from MGF - Clustered MS/MS spectra represented as an MGF file - scan numbers correspond to cluster index in
- CLUSTERSUMMARY PAIRS - Molecular Networking alignment pairs representing spectral similarity
- IDENTIFICATIONS - Spectral library identifications reported by living data/continuous id to the latest GNPS spectral libraries (scan numbers also correspond to cluster index)
Page Contributions¶