Outline
2. Uploading Custom Proteomic Data
2.1 Adding Custom Proteomic Data
2.2 Uploading and comparing Custom Data sets in the Putative Protein Binder table
3. Adding assayed compounds to the Putative Protein Binder table
4. Incorporating Custom Measurements into Ligand Express
4.1 Uploading measurements for annotations in libraries
4.2 Converting measurements into POEM models
1. Batch Deconvolution
Ligand Express’ Batch Deconvolution feature can be used to identify common targets between multiple compounds. First you’ll need to add your compounds to a library and perform Proteome Screens for each of them (please refer to the Getting Started Guide to learn more about running Proteome Screens). You’ll then be able to perform a Batch Deconvolution which combines the results from the Proteome Screens and ranks the proteins based on their shared likelihood of interacting with all submitted compounds.
To perform a Batch Deconvolution, first navigate to a library and select the compounds of interest. Then click the “Batch Deconvolution” button.
In the Batch Deconvolution window, give your Batch Deconvolution a description (optional, but recommended) and click “Yes”. If any of the selected compounds have not yet run Proteome Screens, you’ll be asked if you would like to run them first. Click “Yes” to proceed, then reinitiate the Batch Deconvolution once the Proteome Screen has been completed. Note, as with normal Proteome Screening, you will require Credits (insert credit icon) to run the Proteome Screening jobs. BatchDeconvolution jobs, however, do not require any credits and can be performed as many times as you wish.
Once the calculations are complete, you will receive a notification in Ligand Express. To view the results of your Batch Deconvolution, click on “Batch Deconvolution” under PROTEOMICS in the left-hand sidebar. Then click “Launch” in the “Analysis” column on your experiment’s row.
This will launch a new Proteomics Analysis window or tab with your Batch Deconvolution results.
Proteomics Analysis allows you to view ranked target proteins that are predicted to bind with your compounds, much like you would for a regular Proteome Screen. The “Rank Percentile” column shows the rank percentile of each protein based on the commonality in ranks amongst all compounds from the Batch Deconvolution. Further, each compound has its own column to the right of the table (e.g. column 10w-i and 10w-b) which shows the rank percentile of each protein for that compound only.
You can click on the Rank Percentile column header to sort the table in descending or ascending order by rank percentile. This will allow you to view which targets are most shared amongst your compounds of interest.
2. Uploading Custom Proteomic Data
2.1 Adding Custom Proteomic Data
You can use Ligand Express' Analysis tools to investigate your own data with Custom Proteomic Data uploads. Additionally, Custom Proteomic Data can be added to Cyclica's Proteomic Data to better identify mutually high ranked targets. To learn more about Proteome Screening in Ligand Express, check out Section 6: Submitting Proteome Screening Jobs in the Getting Started Guide. To start using Custom Proteomic Data, click the "Custom Data" link under the PROTEOMICS heading in the left-hand sidebar.
To upload a new set of custom data, click the "Add New Data" button in the top right. In the resulting pop-up, enter a name for your data set and upload the data set file. The file should be in CSV or TSV format and contain protein Uniprot accession numbers and scores. Scores represent the likelihood of interaction with each Protein Target. For interpretability and comparison with Cyclica's Proteomic Data, scores will be converted into Rank Percentiles within the Proteomic Analysis Tool.
Once uploaded, click the "Launch" link next to the newly added proteomic data set to launch it in Analysis view. In this view you can analyze top ranking protein targets and learn more about associated diseases, known binders, tissue abundance, pathways, and more. See the Getting Started Guide for more information on using Analysis view.
2.2 Uploading and comparing Custom Data sets in the Putative Protein Binder table
You can also add Custom Proteomics Data to an existing custom data set, Proteome Screen, or Batch Deconvolution. From the Putative Protein Binder table in Analysis view, click the three vertical circles (⋮) in the top left corner of the table and then click "Upload Custom Data". You will be asked to select a Protein Reference ID (Entry or Accession), upload a CSV or TSV file, and confirm whether or not your file has a header row.
After click "Submit" your Custom data will show up as an additional column in the Putative Protein Binder table.
In this way, you can compare protein target ranks across multiple data sets as well as apply filters to each data set to better identify protein targets of interest.
3. Adding assayed compounds to the Putative Protein Binder table
To add assayed compound information to the Putative Protein Binder table, click the three vertical circles (⋮) in the top left corner of the table and then click "Top Assayed Compound Similarity". This will start running the Top Assayed Compound Similarities and Top Known Binder Similarities analyses for all the known compounds related to each protein displayed in the table. This process may take up to 15 minutes to complete, after which you'll be able to view how similar the screened compound is to the closest known assayed compound or top known binder for each protein. The scores represent the Tanimoto similarity of the screened compound to the top compound known to interact with the protein, where a value of 1 represents an exact Tanimoto similarity.
4. Incorporating Custom Measurements into Ligand Express
4.1 Uploading measurements for annotations in libraries
Custom measurements allow you to upload your own measurements for compound properties and view them alongside Cyclica's ADMET property predictions. Once uploaded, they can be added to the Heatmap view in libraries.
To start using Custom Measurements upload, click the "My Measurements" link under the MODELS & LISTS heading in the left-hand sidebar, then click the "Add new measurement" button.
You can upload a CSV or SDF file with your measurement data by clicking the "Autopopulate From File" button. If the file you uploaded has a header row, select the "Use first row as headers" checkbox. Alternatively, you can start from scratch by typing or pasting your data into the blank spreadsheet provided.
Once you've input your data, select whether the data type is Quantitative or Classification (only one of these types can be input at a time). Then use the dropdowns to define the two columns that contain the SMILES and compound names.
Clicking "Next" will give you an opportunity to correct any errors in the data. After reviewing your data, click "Next" again to enter details on your new measurements. The Measurement Title and Short Name fields will be auto-populated from spreadsheet but can be edited at this time. You can also input a brief description of the measurement's function. Clicking "Submit" will create the measurement. Once the measurement is uploaded, it's details will be viewable on the "My Measurements" page.
Measurements can be added to libraries by clicking the "Heatmap Properties" button on the respective library page, then clicking "Measurements". From here, you can either upload a new measurement via the "Add new data" tab and following the steps detailed above, or you can select an existing measurement to add to the library via the "Choose existing" tab. Then click the "Update" button to add the measurement to heatmap view.
4.2 Converting measurements into POEM models
POEM is Cyclica's ADMET properties prediction algorithm. Custom measurements can be converted into POEM models and used to predict the properties of new compounds. The larger the data set, the more reliable the generated POEM model will be.
To convert a measurement into a POEM model, first go to the "My Measurements" page and select the link to the Details Page for the measurement you'd like to convert. Then click the "Create Model" button. On the following page, select the type of property that the model belongs to from the "Property Group" dropdown and the property label that is associated with the positive case from the "Positive Class" dropdown, then click submit. It may take anywhere from a couple hours to a couple days to generate your model depending on the size of your measurement data set.