Does a Glucose Monitor Help You Lose Weight?
A continuous glucose monitor measures how you eat. It does not cause fat loss. What the data supports, and how to run your own meal experiments.
A continuous glucose monitor measures how you eat. It does not cause fat loss. What the data supports, and how to run your own meal experiments.
A working guide to HMDB's access paths: the licensed REST API, the per-record XML endpoints, and the bulk downloads. Includes a streaming parser, a DuckDB mirror, and mass-based lookup with ppm windows.
There are two products sold as a 'methylation report': an MTHFR-style SNP panel from genotyping data, and a genome-wide DNA methylation array read out as epigenetic clocks. They measure different things, and only one of them is a measurement of your current biology.
Web platforms are good at the downstream half of RNA-seq (differential expression, enrichment, plots) and bad at the upstream half (FASTQ to counts). Here is the split we'd use, with the commands, parameters, and failure modes.
A concrete walkthrough of the RNA-seq workflow from FASTQ to interpreted results: quality control, pseudoalignment with salmon, count import with tximport, differential expression with DESeq2, and the sanity checks that catch most errors.
A practical guide to adapter, quality, poly-G and poly-A trimming for RNA-seq FASTQs: which tools to use, the exact flags, how much trimming is enough, and how to verify you did not damage the data.
A working setup for looking up any SNP in your own sequencing data. Covers indexing the VCF, annotating with rsIDs and verifying genotypes in the reads.
A working pipeline for calling somatic SNVs and indels with GATK Mutect2, from BAM to annotated VCF. Covers results at low variant allele fraction.
A step-by-step guide to annotating a personal WGS VCF with Ensembl VEP. Covers cache setup, the plugin stack and consequence picking.
A practical account of continuous glucose monitor insertion, comfort, accuracy limits, and how to work with the exported data if you are not diabetic and want a metabolic baseline.
A working pipeline for going from RNA-seq FASTQs to junction counts, PSI values, and a shortlist of splice-altering variants from your own genome, with the tools and parameters we would use.
Overnight glucose normally falls to its lowest point of the day a few hours after sleep onset. This page explains what a normal nocturnal curve looks like, how to analyze your own CGM data for real low events, and how to recognize the compression artifacts that mimic them.
Why a post-meal glucose reading in the 80s is usually a healthy result, why you can feel shaky at that number anyway, and how to tell a real reading from sensor noise using your own CGM and lab data.
A working guide to clustering bulk and single-cell RNA-seq: normalization choices that decide your dendrogram, gene-module discovery across timepoints, choosing k defensibly, and the failure modes that produce confident nonsense.
A working guide to the Bioconductor DEP and DEP2 packages for one person's mass-spectrometry proteomics. Covers import, missing values and design.
A practical guide to DNA analysis tools: what works on a 23andMe-style genotype file, what requires FASTQ/BAM/VCF from whole-genome sequencing, and the commands we would run at each step.
A practical guide to wearing a CGM without diabetes. Covers which sensors to buy, how accurate they are and what a normal glucose curve looks like.
A practical guide to Gene Ontology analysis of RNA-seq: building a defensible gene list, correcting for length and expression bias, running overrepresentation tests and GSEA in R, and collapsing redundant terms into something you can read.
A step-by-step guide to going from raw FASTQ files to an aligned CRAM, a benchmarked variant call set, an annotated shortlist, and an RNA-seq expression table, with the actual commands, parameters, and quality thresholds we use.
Read depth in RNA-seq is the number of reads assigned to your sample and, per gene, the number of reads supporting each transcript. This page explains how to compute it from your own files, why 30x means nothing here, and what depth to ask for.