Repository Metadata Database
Note: more important than absolute numbers, emphasis should be put on understanding trends and rates of change. Some fundamental approximates were made, given the lack of information (e.g. flow-cell utilization, academic discount packages, ...)
Taking advantage of public datasets (with particular emphasis on ENA), I have collected sequencing metadata estimated to cover ~95 % of all experiments present in major repositories from 2016 – 2025.
This accounts for ~3 × 10¹⁴ reads, ~30 million runs, ~30 million experiments.
For every accession level - sample, run, experiment, and study, the dataset captures the sequencing platform and model, total read/base counts, detailed library metadata (strategy, source, selection), and the originating center name.
Annual Sequencing Expenditure and Instrument Breakdown
Cost estimations were derived using custom formulas tailored to each sequencing platform and model, combining vendor reagent lists and, wherever possible, adjusted for run-throughput differences and historical price changes.
Below are the key trends and potential areas of technological disruption.
Top: Annual growth in sequencing expenditures, peaking in recent years. Illumina remains the leading platform, but there is a steady rise in MGI / BGI, PacBio, and Oxford Nanopore.
Bottom: Breakdown of total sequencing expenditure by instrument model (cumulative 2016 – 2024), underscoring Illumina’s market dominance.
2023 – 2024 % expenditure changes per platform and model
Experiment Cost and Read Length Across Platforms
Bubble plot summarizing cost-efficiency across sequencing platforms. Each bubble represents an instrument model, positioned by its average cost per gigabase (y-axis, log scale) and total bases sequenced (x-axis). Bubble size reflects usage volume.
High-throughput platforms (e.g., NovaSeq, HiSeq X) appear in the bottom-right quadrant, showing strong cost-efficiency. Older or low-throughput systems (e.g., MiSeq, Genome Analyzer II) show higher per-Gb costs, while long-read platforms (e.g., PacBio Sequel II, PromethION) occupy distinct mid-throughput, higher-cost territory.
Scatterplot showing the relationship between average read length and experiment-level sequencing cost across platforms.
Short-read technologies (Illumina, MGI, ThermoFisher) dominate the low-read-length region (~100 – 300 bp), covering a broad range of costs. Long-read technologies (PacBio, Oxford Nanopore) cluster at longer read lengths and higher cost variability, reflecting their niche in complex or structure-resolving sequencing.