Blog

How 21 CFR Parts 801 and 809 govern device claims and label content, from UDI and prescription statements to IVD package inserts.

A technical walkthrough of the NIH All of Us genomic dataset: sequencing specs, callset formats, Researcher Workbench access tiers, and what a participant receives from their own DNA.

The number of drug and medical device applications China's national regulator accepted and approved in 2025, and what each count measures.

A genetic report is a short curated document about a handful of variants. The data underneath it is a CRAM and a VCF with about five million variants. Here is how to read both, and how to check the report yourself.

A practical software stack that takes a personal 30x whole genome from FASTQ to annotated variants, with the failure modes that waste days.

A direct account of WGS applications, from rare-disease diagnosis to pharmacogenomics, plus the file formats, tools, and flags you need to analyze a 30x genome yourself.

A whole genome array genotypes or copy-number-profiles preselected positions across all 24 chromosomes. Here is what the platforms resolve, the file formats you get, and how the data compares to 30x whole-genome sequencing.

Genome mapping means at least four different things: historical linkage maps, aligning your reads to a reference, optical genome mapping for structural variants, and the colloquial sense of getting your genome sequenced. Here is how to tell them apart, what each costs, and the pipeline we would run on your own data.

How Regulation (EU) 2017/745 sets the clinical evidence for a CE marked device, covering clinical evaluation, investigations, equivalence and PMCF.

A working pipeline for going from a raw germline VCF to a short list of variants you can reason about: normalization, annotation, frequency and evidence filters, pharmacogenomic star alleles, and the failure modes that make most self-interpretation wrong.

What a genome analyzer is at the file level, from FASTQ to CRAM to VCF, and how to measure your own error floor against GIAB.

A homocysteine test measures total plasma homocysteine, a sensitive readout of B12/folate status, renal function, and methylation flux. Here is how to order it so the number is real, what to run alongside it, and how genotype fits in.

How to go from a whole-genome VCF to CPIC-grade star allele diplotypes for CYP2C19, CYP2D6, DPYD, TPMT and others, which callers to run, and the failure modes that make a clean-looking result wrong.

Array-based consumer kits genotype a few hundred thousand preselected sites accurately and everything else badly. If you want data you can analyze yourself, you want 30x whole-genome sequencing with FASTQ, BAM/CRAM, and gVCF returned.

A working pipeline for taking 30x PCR-free FASTQs to aligned CRAM, a filtered and annotated variant set, structural variants, and the QC numbers that tell you whether any of it is trustworthy.

A clinical whole-genome sequencing report is a short curated document about a handful of variants. The FASTQ, CRAM, and VCF behind it contain roughly four to five million. Here is what each one gives you and how to work with the data yourself.

How CLIA certification, performance specification validation and personnel qualifications apply to a laboratory before it reports a patient test result.

A practical guide to loading your own whole-genome alignments, variant calls, and RNA coverage into a local genome browser, indexing the files correctly, and querying annotation programmatically instead of clicking through a web UI.

Hemoglobin is a concentration, not a mass, and that fact explains most of the confusion around a single result. Here is what the assay does, how much of a change is real, and what to pair it with.

MCHC is a computed ratio of hemoglobin to hematocrit, and most high values are analyzer artifacts rather than biology. How to tell the difference using your own CBC history, and the few red cell conditions that genuinely raise it.