The GNPS web interface provides a quick and easy way to perform initial analysis of your data particularly if you want to easily view the MS2 spectra of the nodes/clusters generated by the Molecular Networking workflow.
To access your data, either click the link in the email or click the DONE link under Status on the jobs menu page.
You have several options in the Status bar:
| Data Analysis Option
| View All Library Hits
|| View spectra with library/database matches
| View All Clusters with IDs
|| View all clusters/nodes with library/database matches
| View All Clusters with IDs (Beta)
|| View all clusters/nodes with library/database matches with Experimental Features
| View Network, Node Centric
|| View neighboring nodes to selected node
| View Network Pairs
|| View pairwise network connections between clusters/nodes
|View Spectral Families||List of all Spectral Families (i.e. connected components in the network) as well as view the network visualized in the browser.|
|Networking Statistics||Brief statistics about the network such as number of spectra, number of nodes, number of spectral families, identification rates, rarefaction curves, etc.|
For ease of explanation, only View All Clusters and View Network, Node Centric are highlighted.
View All Clusters allow a user to view the constitutive spectra and the overall average spectrum of a cluster on GNPS.
Click the View All Clusters link.
A list of Clusters will appear. The data can be filtered for a variety of values including PrecursorMZ which is shown here.
To see the average spectrum of the cluster, click the small spectrum icon.
To view the constitutive spectra of the cluster, click the ClusterIndex link.
Again, to view the individual spectra from each file, click the spectrum cartoon.
After data analysis if any of these spectra were manually identified to be a compound that you would like to add to a spectral library, you can see the Gold, Silver, and Bronze links to add to the appropriate quality of spectrum. Further documentation can be found here.
View Spectra Families allows users to look at the molecular networks at a higher level. Users can see all the connected components. These connected components are all the nodes that are connected to each other. From here users can tell how big each component is, how many raw spectra are captured by the particular component, and also the libraries identifications of all the spectra in each component.
The view will look like this:
While this gives users some information, to really dig deeper into the networks, users can click on the component number in the Component column to open a new tab to actually visualize the network in the browser, taking them to the Network Browser View.
An example Network Browser is shown here:
Here users can explore the connected component in the molecular network. Each circle in the left network panel represents a consensus spectrum and edge represents related fragmentation patterns. The default labeling is the cluster/node index which is rather uninformative, but there are alternative labelings possible by selecting the appropriate node label from the Node Label legend box. The description of each item is listed below:
|cluster index||cluster index used globally for each cluster across all networking views|
|parent mass||precursor mass of the consensus spectrum|
|Library ID||library identification name for the node if a library search were performed|
|EvenOdd||Lists 1/0 for even or odd depending on the parent mass. Used for nitrogen rule|
|Peptide||Lists peptide labeling|
In addition to relabeling the nodes, users can also relabel the edges. The table below describes what each label means:
|Cosine||Cosine score measuring fragmentation similarity between two nodes|
|DeltaMZ||The Delta MZ of the spectra|
An additional way to customize the view of the networks is to color the nodes. By default the nodes are gray, but users can color the nodes based upon spectra counts of the groups they came from. By default the coloring labels will be based upon the default groups G1, G2, G3, G4, G5, and G6. The coloring of each node will be a pie, and the proportion colored of that pie is the proportion of the spectral counts coming from each respective group. Additionally, the Node Coloring legend describes the color for each group. Users can also use their own group names if they defined their own arbitrary groups when creating the network.
An example node coloring is as follows:
Edges by default include arrows pointing from low mass spectra to high mass spectra. Further, edges can also be colored. Users are able to enter a delta mz of their choice to highlight in the network into this box:
and hit the circle arrow to update the network. The highlighted edges will appear in red, e.g.:
While its great to visualize the network properly, to truly make sense of the network users must be able to interrogate the network metadata and examine the actual spectra. To this end, users are able to find out information about a particular node by simply hovering their mouse over node. This will bring up a hover box describing the node information. At present the display shows like this:
As we can see, it shows basic information about the node, as well as the structure if the node is identified by library search and the structure was provided in the library spectrum.
To further investigate the network, users can left click on a node to plot the spectrum in the top right panel, along with the spectrum information and structure if one is available:
The spectrum image is interactive so users can zoom in and look at precisely at the spectrum to make sense of the data. Additionally users can compare two spectra in the network by simple right clicking on a different node to render its information in the bottom right panel.
Users can also click on an edge and both nodes that are connected to the edge are automatically displayed in the top and bottom right panels.
Users can quickly ascertain how two spectra align to each other at the peak level. By clicking the "Align Spec" button:
to obtain the score between the spectrum in the top and the spectrum in the bottom render panels.
Additionally, the plots of the spectra will update to reflect the alignments shown here:
with red peaks representing peaks that match at the exact same masses between the top and bottom spectra, and blue peaks representing peaks matching at shifted masses. Additionally, each peak is labelled with a unique identification number that shows the correspondence of peak matches between the top and bottom spectra.
To view how exactly a spectrum in the network has matched to a library spectrum, users can click the "Show Library Match" in the side panel on the far right. This will bring up this middle panel shown below comparing the spectrum in the network with the library spectrum in green.
View Network, Node Centric allows a user to view the spectra of node/cluster neighbors based upon cosine score values.
Click the View Network, Node Centric link. A list of cluster will appear. Click the double arrow to view the neighbors of a node/cluster.
To view the individual spectra from the neighboring nodes, click the spectrum cartoon.
To view PCoA plots, please enable that output on the input page (default is on) in the "Advanced Output Options" section and enabling "Create Cluster Buckets and BioM/PCoA Plots output". This will create the appropriate data structures for PCoA analysis. To include additional metadata to color the nodes according to the sample types, include and add additional columns for metadata describing the experiment (metadata documentation ).
To view the PCoA plots when the parameters are appropriately set, click the "View Emporer PCoA Plot in GNPS" under the section "Advanced Views - Third Party Visualization" on the results page of your GNPS job.