Structure-based prediction of BRAF mutation classes using machine-learning approaches
The BRAF kinase is crucial in oncology because its mutations can activate the MAP kinase pathway, causing cancer
but also enabling targeted therapy. BRAF mutations are classified into three classes (I, II, III), aiding
treatment decisions.
New BRAF kinase mutations often lack classification. To address this, we developed a machine learning tool to
predict the class II and III of BRAF missense variants.
For further information, please have a look at the corresponding
publication.
These results are predictions based on knowledge at the date of the analysis, and are likely to be
refined, completed or corrected over time, depending on scientific and technical advances.
As not all biological aspects of an alteration can be taken into account, these results must be
considered as indicative and, in no case, as a definitive argument for the choice of a
therapeutic strategy.
BRAF Mutation Prediction
Fill the "BRAF mutation request" section with the desired mutation respecting the one or three-letter code.
Example: the BRAF mutation of the Gly469 into Arginine can be written like:
p.G469L
p.Gly469Leu
G469L
Gly469Leu
BRAF mutation request
Links to protein databases
BRAF mutation prediction result
Waiting for a BRAF kinase mutation
3D structure
The structure displayed corresponds to the AlphaFold model of the BRAF protein kinase domain, in active conformation, to ensure that all residues are present.
Controls:
Mouse left-button + drag: rotate the system
Mouse right-button + drag: translate the system
Mouse wheel up and down: zoom and unzoom
Hitting "i" key: roll the system. Hitting "i" again stop the rolling.
Hitting "k" key: rock the system. Hitting "k" again stop the rocking.
A double-click on a residue will center the view on this residue, while displaying it in ball and stick and the close neighbours in licorice.
At the same time, if the double-click has been made on a residue in the main chain, the variant table, and the homologous sequence alignment, will focus on the same residue.
A single-click on a residue will center the system on it, without changing the selection.
Hovering the mouse pointer on an atom will display the atom name, as well as the residue name and number above the 3D viewer.
It is possible to display several types of molecular interactions between the selected residue and its environment using the different options available below the 3D viewer.
Full screen viewing can be obtained by clicking on the .
Hydrogen bonds
Ionic interactions
Cation-π interactions
Hydrophobic contacts
π-stacking interactions
∅
Orthologous sequence alignment
Press "Run with JS"
∅
A single-click on the residue number or a double-click on a residue letter in the sequence alignment will center the 3D structure and the table of variants accordingly.
Residue numbers are displayed above the sequence alignment for the human protein, if they are present in the selected 3D structure.
RAF subfamily sequence alignment
A multiple sequence alignment of protein sequences of the same subfamily of the selected protein. The sequences list and family attribution are based on UniProt database. All sequences retrieved from UniProt data.
Press "Run with JS"
∅
TKL Ser/Thr protein kinase family sequence alignment
A multiple sequence alignment of protein sequences of the same family of the selected protein. The sequences list and family attribution are based on UniProt database. All sequences retrieved from UniProt data.
Press "Run with JS"
∅
Orthologous sequence alignment (extended)
A multiple sequence alignment of orthologous sequences of a protein against human's one. All sequences are retrieved from UniProt database.
.
