Using Bibliometric Analysis and Machine Learning to Identify Compounds Binding to Sialidase-1.

Area of Science:

• Biochemistry
• Computational Biology
• Genetics

Background:

• Rare diseases lack treatments due to limited resources and data.
• Sialidosis is an ultrarare lysosomal storage disorder caused by NEU1 gene mutations, leading to sialidase-1 deficiency and toxic buildup.
• Current treatments for sialidosis are unavailable.

Purpose of the Study:

• To develop a novel approach for rare disease drug discovery using bibliometric and machine learning tools.
• To identify potential drug candidates for sialidosis by screening compounds against sialidase-1.
• To validate the efficacy of identified compounds through in vitro testing.

Main Methods:

• Utilized bibliometric tools to link lysosomal storage disease targets with existing bioactivity data.
• Curated literature data to build a Bayesian machine learning model for compound screening.
• Employed in silico screening and in vitro microscale thermophoresis for compound validation.

Main Results:

• Developed a machine learning model to screen compound libraries for sialidase-1 binding.
• Identified two compounds, sulfameter (Kd 2.15 ± 1.02 μM) and mexenone (Kd 8.88 ± 4.02 μM), with significant binding affinity.
• Validated the computational approach for identifying novel molecules with therapeutic potential.

Conclusions:

• The combined bibliometric and machine learning approach effectively aids in small molecule data curation and model building for rare disease drug discovery.
• Sulfameter and mexenone show promise as potential drug candidates and chaperones for sialidosis.
• This strategy can accelerate the identification of novel therapeutics for ultrarare diseases with no current treatment options.