The problem. A modern bioinformatics run scatters quality metrics across dozens of files: FastQC per sample, alignment stats from the aligner, duplication rates, counts summaries. Inspecting them one by one, per sample, doesn’t scale — and problems (a bad lane, an outlier sample) hide in the pile.

The idea. MultiQC crawls an analysis directory, recognises the output formats of a long list of tools, and aggregates their metrics into a single interactive HTML report with per-sample plots. You see all samples side by side for every metric at once, so outliers jump out immediately.

Why it matters. This one isn’t abstract — MultiQC is the reporting step in my nf-core-style aml_rnaseq_nf and variant_calling_nf pipelines. Reading the paper explains why it’s a near-universal default: it standardises the “did this run go okay?” question across tools, which is exactly the reproducibility discipline the pipelines are built for. It’s small, but it’s the QC habit made frictionless.

Verdict. Foundational to practical workflow hygiene and ubiquitous in nf-core. Not a method so much as glue — but the right glue, and the reason I catch a bad sample before it reaches DESeq2 or GATK. Read it for how much friction good aggregation removes.