Blog

What structural variants are, which callers to run on short-read and long-read whole-genome data, the flags that matter, and how to tell a real call from an alignment artifact.

A working guide to variant effect predictors: what they score, how to run VEP with the right plugins, and which scores to trust.

A working pipeline for searching the 4 to 5 million variants in your genome: annotate, filter to a few dozen candidates, then check each by hand.

How FDA premarket routes divide between 510(k), De Novo and PMA, with the criteria, contents and review timeframes set by 21 CFR parts 807 and 814.

6 mmol/L is 108 mg/dL, and an A1C of 6.0% is a different number entirely. Here is how to tell which one you have, why a single reading carries little information, and how to get a real answer from CGM and lab data.

A fasting glucose of 99 mg/dL is one number from one morning. Here is how to compute your own morning baseline from CGM data, what the dispersion means, and where the measurement itself is the limiting factor.

A specific panel to start from, the pre-analytical variables that corrupt a longitudinal series, and how to decide whether a change between two draws is real.

A functional biomarker measures how well a system is working — a ratio, a rate, or a response to a challenge — rather than a single resting concentration. Here is how to build them from CGM, blood panels, proteomics, and RNA-seq, and where they break.

A working pipeline for a personal whole genome: alignment, variant calling, annotation, and the QC steps that catch the errors most people miss. Tool names, flags, file formats, and failure modes.

A working pipeline that turns a personal whole-genome VCF into an annotated variant table you can query, using VEP, gnomAD and ClinVar.

A working list of the tools that matter for analyzing your own genome, transcriptome, proteome, and biomarker data, with formats, flags, and the failure modes that bite.

What fasting, post-meal, and 24-hour continuous glucose numbers look like in people without diabetes, what the reference data say, and how to analyze your own CGM export.

A working pipeline for whole-exome data from FASTQ to VCF, with the filters that matter and the places where exome capture loses to a whole genome.

What whole exome sequencing covers, where capture fails, and how to check the quality of an exome you already own.

A working pipeline from a counts matrix to a ranked, deduplicated pathway table for personal RNA-seq, including the n=1 designs that are valid, the blood cell composition problem, and which tools to trust.

How FDA assigns devices to class I, II or III under 21 CFR Part 860, covering classification regulations, De Novo requests and 510(k) exemptions.

A practical guide to comparing whole genomes: your VCF against the reference, two callsets of the same sample, your genome against a relative's, and human against other species. Tools, flags, and the failure modes that produce fake differences.

What pharmacogenomic panels like GeneSight measure, what the trial data show, how to read CYP2D6 and CYP2C19 from your own genome, and what serum drug levels can and cannot tell you.

What the five ACMG tiers mean on a BRCA report, how to pull the same calls from your own VCF, and what short-read sequencing misses.

What CGM error looks like in the data: interstitial lag, compression artifacts, sensor-to-sensor bias, and a protocol for measuring your own sensor's accuracy against fingersticks using the raw exports.