Differential Gene Expression Analysis in R with DESeq2
A working tutorial for taking RNA-seq quantifications into R, fitting a negative binomial model with DESeq2, and reading the output without fooling yourself.
A working tutorial for taking RNA-seq quantifications into R, fitting a negative binomial model with DESeq2, and reading the output without fooling yourself.
A working pipeline from FASTQ files to a ranked table of differentially expressed genes, including longitudinal sampling of one person.
Sensor error during exercise is large relative to the glucose range of a non-diabetic athlete. Here is what MARD means in practice, which failure modes dominate during training, and how to design and analyze a CGM experiment that survives the noise.
A practical walkthrough of a bulk RNA-seq count matrix: where the numbers come from, how to load and normalize them, and how to tell signal from artifact in your own data.
A technical read on Dante Labs' consumer whole-genome sequencing: the files they deliver, how to verify coverage and contamination yourself, and how to re-call and annotate variants from the raw data.
A practical guide to pulling the dbSNP VCF from NCBI, verifying it, fixing contig names, and using it to annotate your own variant calls.
DeepVariant is a deep learning germline variant caller that turns aligned sequencing reads into a VCF of SNVs and small indels. This page covers what it does, how to run it, how to read its output, and how to measure its accuracy on your own data.
baseMean is the average of normalized counts across all samples. Thresholding on it after results() is the wrong move; filter raw counts before fitting and let independent filtering do the rest.
How 21 CFR Parts 201 and 202 govern drug and biologic labels, prescribing information, Drug Facts panels and permitted claims in advertising.
Most adults without diabetes spend the great majority of the day between 3.9 and 7.8 mmol/L with a mean near 5.0 to 5.5 mmol/L. This page gives the reference numbers, the sensor error that sits underneath them, and the code to compute your own metrics from raw CGM exports.
Healthy adults wearing a continuous glucose monitor spend most of the day between 70 and 140 mg/dL with a mean near 100. Here is where those numbers come from, what sensor error does to them, and how to compute your own metrics from the raw export.
A price-per-day comparison of over-the-counter continuous glucose monitors in the US, what each one gives you in terms of data access, and how to get raw glucose values out of them.
A practical guide to calling germline small variants with Clair3 from ONT or PacBio HiFi data, with model selection, phasing and benchmarking.
A step-by-step guide to going from FASTQ files to sample clusters and gene co-expression patterns in your own transcriptome data, including which distances to use, how to choose k, and how far the interpretation can reasonably go.
A practical guide to the shell tools, file formats, and indexing habits you need to query your own genome, transcriptome, proteomics, and glucose data without a GUI.
A practical Python stack for querying your own whole-genome and RNA-seq data: cyvcf2, pyranges, pysam, polars, and the coordinate conventions that quietly break analyses.
How Directive 2001/83/EC sets the evidence for a marketing authorisation, covering full dossiers and the generic, biosimilar and well-established use routes.
Dehydration concentrates the blood a little, but it does not explain most elevated AST or ALT results. Here is the size of the effect, the confounders that matter more, and how to retest so the number means something.
Endurance training raises AST and sometimes ALT, because skeletal muscle contains both enzymes. How to tell muscle from liver in your own lab data.
Why an HbA1c can read high while daily glucose looks normal, how to tell assay interference from red cell biology, and how to confirm the mismatch with CGM, fructosamine, and your own genome.