Targetdecoy

Targetdecoy, more commonly known as the target-decoy strategy, is a method used in mass spectrometry–based proteomics to estimate the false discovery rate (FDR) of peptide-spectrum matches. It uses a decoy database of artificial sequences in addition to the real target sequences, providing a null model for random matches. By searching spectra against a concatenated target-decoy database, identifications that map to decoy sequences help gauge how many incorrect identifications are expected, enabling calibration of confidence scores and FDR control.

In practice, decoy sequences are generated by reversing, shuffling, or otherwise permuting real protein sequences. The target and decoy databases are combined, and spectra are searched against this mixed database. Each match receives a score, and the numbers of decoy and target identifications above a score threshold are used to estimate FDR, often with FDR = decoy matches / target matches, or a related variant. A threshold is chosen so that the estimated FDR meets a desired level, such as 1% at the peptide-spectrum match (PSM) level. The approach is also extended to peptide- and protein-level FDR calculations, sometimes with additional validation steps.

Targetdecoy is widely integrated into proteomics workflows and software tools (for example, in popular database search engines and rescoring pipelines). It rests on assumptions about decoy similarity to true false positives and the balance of decoy to target sequences; its effectiveness can depend on decoy design and the data set. Despite limitations, the target-decoy strategy remains a standard method for FDR estimation in peptide identification.