Table Builder Tiered Filtering Explained

The role of any filter is to narrow down the results in a table to help you hone in on the data that is relevant to your research or exploration. Filters operate by adjusting which rows of data are included so that the column you are filtering aligns with your specifications. 

An example of this would be filtering a Drugs table with a Drug Name filter. For example, if you were only interested in exploring Acetaminophen you could add a Drug Name filter, as seen below. The table results would only include rows of data where the Drug Name matches Acetaminophen. 

Tiered Filtering

While most filters apply to the entire table, there are a handful of exceptions. Specifically, you will notice that filters found under Related Drugs, Related Proteins, and Target Disease Associations all operate slightly differently. When filters are applied to these columns, the filter first narrows the data within that subset, then refines the overall table results. This is an expected and intentional operation that we refer to as tiered filtering. 

In the example below we have a Proteins table that includes Related Drugs. Most filters you select in a Proteins Table will start by narrowing the Proteins data, seen here in the pink box. Any Related Drugs filter you select will first narrow the Related Drugs data, within the blue section, before impacting the Proteins data. 

In this example we’ll start with a Proteins filter. If you were only interested in exploring the Genes doxG and mdlB, you could add a filter for Gene Name, seen below.

The result of this filter will be the protein (S)-2-haloacid dehalogenase being removed from the table, as it doesn’t include either of the selected Genes. 

Once the filter is applied, the only Proteins you would see are those related to the Genes doxG and mdlB:

Now, let’s shift focus and look at the Related Drugs section of this table. Our filtering expectations will change slightly. Any Related Drugs filter you select will first act on the columns included under the Related Drugs heading, then act on the Proteins data. 

An example of this would be if you wanted to exclude any Related Drugs that had an Experimental Approval Status. 

We would add the filter: 

This filter would first act on the Related Drugs columns of data, removing any Drugs that have an Approval Status of Experimental. Below, we can see the pink boxes, which are the lines of data that will be excluded. 

In the case of the proteins (S)-mandelate dehydrogenase and (S)-2-haloacid dehalogenase, each has a Related Drug with an Approval Status of either Approved or Approved, Investigational so both will remain in the table results after the filter is applied. 

The protein 1,2-dihydroxynaphthalene dioxygenase, outlined in blue, only has Related Drugs with an Approval Status of Experimental. As a result, once the Related Drugs: 3,4-Biphenyldiol, Biphenyl-2,3-Diol, and 1,2-Dihydroxynaphthalene are filtered out of the table, the Protein will also be removed. 

After the filter is applied, you would see the following results:

If you are unsure of how the filters are operating, a good indication that you are using tiered filtering is the double layer of headings within a table. All of the columns associated with Related Drugs, Related Proteins, and Target Disease Associations will be nested under a higher level heading within the table. 

Here we can see Related Proteins in a Drugs table. 

And, here we can see Target Disease Associations in a Proteins table. 

Each starting point, including Proteins, Drugs, Clinical Trials, and Diseases, have filters that operate on the table level, and on the secondary level. 

There are instances when it would be helpful to be able to control which level a filter applies to. As an example, you might be interested in uncovering Proteins that currently have no approved Drugs. Seen below, this would look like removing the Proteins (s)-mandelate dehydrogenase and (s)-2-haloacid dehalogenase. 

As our filters currently operate, you would not be able to specify that you wanted a Related Drugs filter to remove a Protein if there is even a single approved Related Drug. 

At this time, the Related Drugs, Related Proteins, and Target Disease Associations filters all only operate on their columns of data up until they have been reduced enough to narrow the higher level table data. We are currently working to create more flexible filter options that will allow you to specify the level of data that is impacted by your filters.

Learn more about the current limitations of tiered filtering and our ongoing work to improve them. 

Got a minute to help us improve? 

We value your experience and are eager to hear your thoughts. Whether it’s a hiccup or a suggestion, your feedback is a vital part of this development process. Send us feedback.