What changed when 21 CFR 820 became the QMSR
How 21 CFR part 820 became the Quality Management System Regulation, which sections were removed or reserved, and where each requirement now sits.
How 21 CFR part 820 became the Quality Management System Regulation, which sections were removed or reserved, and where each requirement now sits.
A practical guide to QUAL, hard filters, and genotype masking with bcftools filter and bcftools view, including the failure modes that silently drop variants from your own genome.
A working guide to bcftools mpileup and bcftools call on a personal 30x whole-genome BAM, including filtering, normalization, BCF versus VCF, and where this pipeline will mislead you.
A precise account of the Benjamini-Hochberg step-up procedure: the arithmetic, the assumptions, the code, and the failure modes that matter when you run it on your own RNA-seq or proteomics data.
Upload-a-PDF analyzers mostly do OCR and template text. A better setup is structured lab data with units and LOINC codes, your own longitudinal baseline, and a general reasoning model you drive yourself.
A practical answer for people analyzing their own molecular data: use a frontier LLM as a coding agent over your files, use task-specific models for prediction, and verify everything at the file level.
How to choose an at-home C-reactive protein test, why the high-sensitivity assay matters, and how to analyze repeated CRP measurements without fooling yourself with biological noise.
What a manufacturer files under Regulation (EU) 2017/745: Annex II and III technical documentation, the route by class and how notified bodies review it.
A curated list of blood, genomic, transcriptomic, proteomic, and glucose markers that carry real information about one person, with the variability numbers and file formats you need to work with them.
A step-by-step method for deciding whether a failing FastQC module means your library is broken, your sequencer had a bad run, or the plot is simply behaving as designed.
A field-by-field guide to the eleven mandatory SAM/BAM columns, the optional tags that matter, and the commands to read, filter, and visualize your own alignments.
An annotated alignment record from a 30x human whole-genome BAM, plus the samtools commands we use to inspect a new file and the failure modes worth checking first.
A practical guide to opening and inspecting BAM alignment files: samtools on the command line, IGV for visual review, and the checks worth running before you trust a variant call.
A baseline blood test is only useful if it establishes your personal set point under controlled conditions and lands in a file you can compute on. Here is the panel we would order, the preanalytical controls that matter, and the math for deciding when a later change is real.
When an IND is required under 21 CFR Part 312, the exemptions, and the informed consent and investigator duties that govern FDA-regulated drug trials.
30x means an average of 30 reads covering each base of your genome, which is enough for confident germline SNV and small-indel calling across roughly 90-95% of the reference. Here is the coverage math, the files you should insist on, and the QC steps we would run before trusting a single variant.
A working guide to RNA-seq analysis for one person's data: library choices, depth, QC thresholds, and how to analyze repeated samples.
How to get a useful AI-generated report from your own DNA data, which file formats matter, and where the interpretation breaks down.
A practical guide to DeepMind's AlphaGenome model: what it outputs, how to call the API on your own variants, and how to check its predictions.
A practical guide to running ANNOVAR on a personal whole-genome VCF: which databases to download, the exact table_annovar.pl command, how to normalize input, and how transcript choice changes the answer.