DIA-NN 2.6: DDA quantification accuracy

  • Highest precursor coverage and data completeness on the Proteobench DDA benchmark.
  • MS1-level quality filtering enables precise control over quantification accuracy, yielding the lowest deviation from expected species ratios.

Quantification Metrics

DIA-NN 2.6 DIA-NN 2.6 filtered Software A Software B
Precursor ratios 94,152 16,620 38,756 15,771
Complete profiles 65,208 12,338 22,149 11,216
Data completeness 80.3% 87.9% 75.7% 75.9%
Human MAD 0.272 / 71,549 0.135 / 13,125 0.215 / 30,199 0.170 / 12,578
E. coli MAD 0.713 / 4,454 0.164 / 512 0.258 / 1,294 0.170 / 346
Yeast MAD 0.373 / 18,149 0.165 / 2,983 0.226 / 7,263 0.177 / 2,847

Precursor-Level Ratio Distributions

Log2(A:B) precursor ratio density estimates for a three-species mixture (ProteoBench DDA benchmark). Dashed lines: expected ratios. Human 1:1, Yeast 2:1, E. coli 1:4.
DIA-NN 2.6 -4 -3 -2 -1 0 1 2 3 4 log₂(A:B) 71,549 4,454 18,149
DIA-NN 2.6, filtered -4 -3 -2 -1 0 1 2 3 4 log₂(A:B) 13,125 512 2,983
Software A -4 -3 -2 -1 0 1 2 3 4 log₂(A:B) 30,199 1,294 7,263
Software B -4 -3 -2 -1 0 1 2 3 4 log₂(A:B) 12,578 346 2,847
HumanE. coliYeast
Experimental notes: Data: ProteoBench DDA quantification module (Van Puyvelde et al., 2022). Three-species mixture (H. sapiens, S. cerevisiae, E. coli), 50 ng, 15 min gradient, 50 cm μPAC column, DDA (180k MS1, 2 Th, HCD). 6 runs (3×A, 3×B). DIA-NN 2.6.0: library-free mode, Precursor.Quantity (non-normalised). Standard: 1% precursor and global FDR. Filtered: additionally Global.Ms1.Q.Value < 0.01, Ms1.Q.Value < 0.01, Global.Ms1.Quality > 0.8. Software A and B: processing details as per ProteoBench public submissions. MAD = median absolute deviation from expected log2 ratio.
DIA-NN 2.6 release notes
For a complete list of new features and improvements, see the full release notes.
10 Jun 2026
Privacy Policy and Legal Notice © 2026 Aptila Biotech GmbH