DC Biosciences is proud to start offering DIA analysis, the latest standard in LC-MS/MS acquisition method. Deep coverage of multiple samples has never been this easy!
Classic LC-MS/MS based proteomics experiments, such as most of the services we have offered until recently, use a Data Dependent Acquisition method (DDA), called “top N”. This means that during each duty cycle, the instrument decides on the fly which are the top N (typically N = 10-15) most interesting peaks and attempts to fragment them one after the other. However, this means that in two different but related samples identifications will be very context-dependent. As a result, DDA data matrices tend to contain missing values where peptides have been identified in some samples only – when, most of the time, they should actually be present in all! This means that on the image here to the right, for label-free DDA data it is very difficult to see whether or what is unusual about the picture.
There are two ways to mitigate this within a DDA framework. The first is to use a method to propagate IDs between samples, such as the Match-between-Runs option in MaxQuant. However, this method does not fill all the gaps, and it generates errors. The second is to label the samples, e.g. with SILAC or isobaric tags (TMT, iTRAQ). In labelled experiments, any peptide identified is quantified in all labelling channels. As of 2017, the highest commercially available multiplexity is 11-plex (TMT). This is sufficient for many experiments, but still means that for experiments with more samples several TMT samples need to be run – and many peptides will only be identified in one set. On the image here, for labelled DDA the picture is a little bit clearer but is is still very random… wait, what does he have on his face?
In recent years, a new family of methods called Data Independent Acquisition (DIA) have really been picking up momentum. DIA methods have the particularity that, for each duty cycle, all precursors are being fragmented. This results in extremely complex, hybrid MS2 spectra, which are then searched using a spectral library to generate an extremely complete, unbiased picture of the samples’ proteomes. In addition, because in DIA all peptides present are fragmented, DIA data can be re-searched later with an improved library to generate even better data! As our libraries improve, we can finally see that the man did, strangely, have an apple stuck on his face.
Originally, spectral libraries for DIA had to be derived from DDA-runs, which were not always available (e.g. when starting work on an organism for which no previous DDA data was available). However, it is now possible to deconvolve MS2 DIA spectra into individual, pseudo-MS2 spectra, which can be searched using conventional algorithms to then generate a library. These recent developments mean that DIA is quickly becoming the new standard in Proteomics research.
When should I run a DIA rather than a DDA method?
Use DIA when you have a large number of samples and when you need to obtain an unbiased, deep picture of their proteomes. Importantly, since DIA is strictly incompatible with isobaric labelling and currently incompatible with SILAC (current DIA algorithms have not been designed for SILAC), it tends to be more expensive, as it does not benefit from the reduction in sample runs afforded by labelling: in DIA, each biological sample will be run independently. However, many DDA experiments requiring deep-proteome coverage have to resort to offline fractionation. By contrast, the use of a library, by allowing deeper proteome coverage, means that DIA can do away with offline fractionation in most cases while still retaining deep coverage.