HomeGenomics

Your genome is yours. Get comfortable reading and understanding it, one step at a time.

What is this site for?

HomeGenomics is a practical handbook that helps developers read and understand their own and their family’s genomic and transcriptomic data. It looks beyond the summary report from a testing provider to examine how raw data is produced and how evidence supports an interpretation.

Now that AI agents can handle execution, memorizing commands matters less than being able to understand the underlying principles, specify what to examine, and interpret both the meaning and the limits of a result. HomeGenomics practices an approach in which the agent handles execution while the reader owns the questions and judgment.

The starting point is a real case. GitLab co-founder Sid Sijbrandij published his rare-cancer data and documented how he worked with specialists to narrow down treatment candidates. We follow that case to learn how to find a signal in data, check it against other evidence, and distinguish a plausible interpretation from its limits.

The published learning paths now cover bulk RNA-seq and single-cell RNA-seq. They connect concepts from library preparation and FASTQ processing to expression matrices, quality control, interpretation, and comparisons across samples. DNA variant analysis is planned next.

Learn from real cases

Using Sid’s published material and public research datasets, we examine how an analysis question is formed between real data and a possible treatment candidate.

Let the agent execute

Step-by-step prompts delegate environment setup, CLI commands, and Python execution. The reader defines the inputs, comparison, outputs, and validation checks, then reviews the result.

Understand it with your family

Relatives share some genetic information. We learn to separate inherited variants from tumor-acquired changes and to confirm family-relevant findings through genetic counseling and validated testing.

Study the limits too

We examine data quality, comparison choices, and uncertainty, then separate supported conclusions from questions that still require validation.