An Experiment on Ab Initio Discovery of Biological Knowledge from scRNA-Seq Data Using Machine Learning

Area of Science:

• Computational Biology
• Developmental Biology
• Machine Learning

Background:

• High-throughput biological data analysis often relies on existing knowledge.
• The capacity of machine learning (ML) for knowledge discovery without prior guidance is under-explored.

Purpose of the Study:

• To investigate the potential and constraints of ML for *ab initio* knowledge discovery in biological data.
• To apply ML methods to single-cell RNA sequencing data from early human embryonic development.

Main Methods:

• Systematic experiments using combined unsupervised and supervised ML approaches.
• Analysis of single-cell RNA sequencing data to identify patterns without pre-existing knowledge constraints.

Main Results:

• A strategy integrating unsupervised and supervised ML successfully identified major cell lineages with minimal human intervention.
• The *ab initio* ML approach led to the novel discovery of human early embryonic cell differentiation pathways.

Conclusions:

• Demonstrates the feasibility and significance of *ab initio* ML for uncovering complex biological insights.
• Highlights the potential and limitations of using ML for knowledge discovery in developmental biology.