KinomeSelector is a tool that helps selecting the minimal set of kinases in inhibitor studies without loosing coverage. The underlying data pooles multiple kinase inhibitor studies to calculate the relationship between sequence similarity of kinases and their inhibitor profile similarities.

Under the assumption that kinases with similar domain sequences are inhibited by the same inhibitors, users can generate their own kinase panels using the interface in a simple two step process.

Basic usage

Step 1: Select an appropriate sequence similarity cut-off. The higher the selected cut-off, the more similar two kinase domains have to be to be considered inhibited by tge same substance.
Step 2: Select the kinases until a full coverage is given. Pink marks the selected kinases while yellow shows the kinases covered by similarity.

In addition, we provide two pre-computed sets of kinase panels with different similarity cut-offs.

Determining the pre-computed kinase screening sets

The algorithm is iterative and consists of 3 steps: A - sorting, B - assigning, C - removing.

  1. The kinases are sorted based on number of other kinases above given cut-off (neighbors)
  2. The protein with the highest number of neighbors is assigned for screening
  3. All its neighbors are removed from the set and sorting (A) is repeated for the pruned set
These three steps are reiterated until all the proteins are assigned to screening or removed from the set. After each step the screening coverage is calculated by dividing the number of kinases assigned and removed by the total number of kinases in an initial set. Kinase domain similarities were obtained by running Smith-Waterman search (ssearch v36.3.5a) with the default settings. Instead of percent identity the alignment bit score was chosen as a similarity measure to avoid using alignment length cut-off.