De-novo protein function prediction using DNA binding and RNA binding proteins as a test case

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• Most identified protein sequences lack experimental functional data.
• Current function prediction relies on homology, leading to inaccuracies and gaps.
• Many proteins lack annotated homologs, hindering functional prediction.

Purpose of the Study:

• To develop a de novo protein function prediction approach.
• To identify proteins with novel functions, such as DNA and RNA binding.
• To experimentally validate predictions made by the new method.

Main Methods:

• Developed a de novo approach identifying biophysical features underlying protein function.
• Applied the method to discover proteins not identifiable by homology.
• Experimentally validated predictions, including DNA binding and subcellular localization.

Main Results:

• Discovered novel DNA and RNA binding proteins missed by homology-based methods.
• Experimentally confirmed FGF14, a secreted growth factor family member, binds DNA.
• Demonstrated that mutated binding sites on FGF14 abolish DNA binding, confirming specificity.

Conclusions:

• The proposed de novo approach enables accurate protein function prediction.
• This method successfully identifies proteins with functions undetectable via homology.
• Automated de novo prediction based on biophysical features is feasible and experimentally validated.