A pipeline is a set of unique tools that can be run sequentially by using the result of one tool as input to the next tool. This allows for seamless integration between different tools within the NXG Tools application. Pipelines can be saved so that they can be shared and re-used with collaborators.

Building a Pipeline

Pipelines are built in a step-by-step fashion. Once the tools and parameters are selected to your liking, the pipeline may be run on many different input datasets. It is recommended to build a pipeline up with a small subset of your dataset before scrolling back to the top and entering your full dataset.

NOTE: There are a limited set of fields from any given step that may be used as inputs for downstream steps. The full list of valid next steps is always available at the valid_steps API endpoint.

In order to build a pipeline:

  1. Configure and run tool 1. Using a small input dataset, adjust the parameters for the initial tool accordingly and click Run.

  2. Apply filters (optional). Using the controls in the Results table header, filter for the rows that you would like to send to the next tool. See the section on Passing Filtered Results To Downstream Tool below.

  3. Select the next tool. When tool 1 results appear, the + button also appears in the Pipeline Map (sidebar). The + button is a dropdown that contains list of tools that are able to use the results from the tool 1 as inputs to tool 2.

  4. Select the column that will be sent. The tool dropdown mentioned above contain submenus that list the columns that can be sent to the next tool. Make the appropriate selection.

  5. Repeat until complete. Repeat steps 1 through 4 until the steps of the pipeline are defined to your need.

  6. Run!. Click Run for each individual step, Run All Above from the Run dropdown on one of the downstream steps.

Run All Above

NOTE: Only the most recently added tool can be removed from the pipeline using the `x` button next to the name of the last under 'Steps'.

Passing Filtered Results To Downstream Tool

  1. After a table column is filtered, the result table is highlighted in yellow and the ‘Save Table State’ button is enabled.

  2. Click on the ‘Save Table State’. The table will un-highlight, ‘Save Table State’ button becomes disabled, and the filtered columns are now saved to the pipeline.

  3. Only the data that was last saved by ‘Save Table State’ will be passed down to the downstream tool.

  4. In order to pass all data downstream, clear all filters and click on ‘Save Table State’. Then select and run the next tool.

Pipeline Map


The pipeline map is a sidebar feature that is used to build and visualize pipelines. It also allows for sharing of complete pipelines, both with and without data for reproducibility. This is explained in Saving/Sharing Pipelines

  • << button - Minimizes the pipeline map

  • pipeline-map-share-button.png dropdown that contains pipeline_id and pipeline_spec_idurls explained in Saving/Sharing Pipelines.

  • Info - Contains inputs for Pipeline Name and Email

  • Steps

    • Each step of the pipeline is listed here, with the name of tool (e.g., T Cell Prediction - Class I).

    • Tool Parameters - Expanding the step, by clicking on the Carrot Icon will reveal the current input values for the selected tool.

    • + button - Add new tool to pipeline. The output of the previous tool, in this case T Cell Prediction - Class I, will be used as the input for the next tool.

    • Icons

      • loading-icon.png Running tool and waiting for results.

      • results-icon.png Results are displayed.

      • filter-icon.png Filters are applied on the result.

Saving/Sharing Pipelines

After a tool runs in the pipeline, the pipeline is saved in the database and can be retrieved using the ‘Link to this pipeline’ (pipeline_id) or ‘Link to pipeline configuration’ (pipeline_spec_id).


Pipeline steps with data (pipeline_id)



The pipeline url contains pipeline_id that is used to retrieve the tools, parameter selections, input data, and results for the pipeline.

Pipeline steps without data (pipeline_spec_id)



The pipeline spec url contains pipeline_spec_id that is used to retrieve only the tools and parameter selections for the pipeline. No input sequences or results are stored with pipeline_spec.

NOTE: Anyone with the link to the specific pipeline or pipeline_spec may retrieve the results without authentication.Pipelines with and without data are currently stored indefinitely on the NXG Tools system. In an upcoming release, a more formal retention policy will be established.

Notes on the Pipeline Lifecycle

  • Run

    • When the Run button is clicked in a step that has previously been run, downstream results are deleted from the page, after a popup confirmation. In the popup, you will be given the URL where you can recover the previous results if needed.

  • Run All Above

    • This selection is available from Run buttons of downstream steps. It will run all the steps up to and including the step where Run All Above was clicked.

  • Cancel

    • The Run button morphs into a Cancel button while a step is running Clicking here will cancel the current execution.

A pipeline ID is created and saved whenever a tool is run via Run or Run All, and when table state is saved via Save Table State.

As downstream steps are dependent upon their input from upstream steps, results from downstream steps will disappear when an upstream step is rerun. A verification modal shows up before this occurs:


Example pipelines

1-Step Pipeline

To start a pipeline, please select a tool from the home page or on the ‘Tools’ dropdown at the top right.

In this example, the 1-Step Pipeline is T Cell Prediction - Class I.

Set parameters and click ‘Run’ to get results.


2-Step Pipeline

To add a 2nd tool to the pipeline, click on the + button and a dropdown will appear.


In this example, the 2-Step pipeline is T Cell Prediction - Class I and Cluster. Because ‘Peptide’ is the only option in this example, the ‘Peptide’ column for T Cell Prediction results will be passed as input sequences to Cluster.

Set parameters and click ‘Run’ to get results.


3-Step Pipeline

To add a 3rd tool to the pipeline, click on the + button and the dropdown will appear. All tools that are already part of the pipeline cannot be selected again (i.e. T Cell Prediction - Class I).


In this example, the 3-Step pipeline is T Cell Prediction - Class I, Cluster, and Pepmatch. ‘Peptide’ and ‘Cluster Consensus’ are both columns that can be passed as input sequences to Pepmatch.

Set parameters and click ‘Run’ to get results.