Comparison of Protein Quantification in a Complex Background by DIA and TMT Workflows with Fixed Instrument Time

Abstract

Label-free quantification (LFQ) and isobaric labeling quantification (ILQ) are among the most popular protein quantification workflows in discovery proteomics. Here, we compared the TMT SPS/MS3 10-plex workflow to a label free single shot data-independent acquisition (DIA) workflow on a controlled sample set. The sample set consisted of ten samples derived from 10 biological replicates of mouse cerebelli spiked with the UPS2 protein standard in five different concentrations. For a fair comparison, we matched the instrument time for the two workflows. The LC–MS data were acquired at two facilities to assess interlaboratory reproducibility. Both methods resulted in a high proteome coverage (>5000 proteins) with low missing values on protein level (<2%). The TMT workflow led to 15–20% more identified proteins and a slightly better quantitative precision, whereas the quantitative accuracy was better for the DIA method. The quantitative performance was benchmarked by the number of true positives (UPS2 proteins) within the top 100 candidates. TMT and DIA showed a similar performance. The quantitative performance of the DIA data stayed in a similar range when searching the spectra against a fasta database directly, instead of using a project-specific library. Our experiments also demonstrated that both workflows are readily transferrable between facilities.

Surpassing 10 000 identified and quantified proteins in a single run by optimizing current LC-MS instrumentation and data analysis strategy

“…In this commonly used strategy for DIA experiments, sample-specific proteins could be potentially missed, because of dilution of these proteins in the pooled sample. This effect has been described previously for a protein spike-in experiment in complex background 30 and will likely get more prominent the more samples are pooled to generate the project-specific library. The largest improvement in quantified precursors (+27%), peptides (+12%) and proteins (+10%) using the hybrid library approach was found for the resource library which was based on DDA data from unrelated samples and on a different LC-MS setup.…”

“…Compared to library-based analysis of the DIA data, a database search resulted in less identified proteins, but an improved quantitative performance. 30 A marriage of both of these approaches, i.e. combining data from a database search of DIA data with results of a search of DDA, holds the potential to the improve DIA data analysis in terms of quantitative precision and number of quantified proteins.…”

“…This was already shown previously, but on the other hand the DirectDIA analysis proved to provide a better quantitative performance than the library-based analysis. 30 To combine the benefits of both approaches, we developed a hybrid library approach.…”

Combining Precursor and Fragment Information for Improved Detection of Differential Abundance in Data Independent Acquisition

“…Specifically, one set of controlled mixtures had defined spike-in of few proteins in a constant background (Spike-in-HEK293-OT dataset). Additionally, a set of controlled mixtures had defined spike-in of few proteins in a background with realistic biological variation (Spikein-biol-var-OT) (23). The third type consisted of sets of controlled mixtures at defined ratios (MP-LFC-OT, MP-SFC-OT, MP-LFC-TTOF, and MP-LFC-MS1var-OT).…”

Enzymatic dissociation induces transcriptional and proteotype bias in brain cell populations

“…Peptides were loaded onto the column with 100% buffer A (99% H2O, 0.1% formic acid) and eluted with increasing buffer B (80% acetonitrile, 0.1% formic acid) over a nonlinear gradient for 120min. The DIA method (Muntel et al, 2019) contained 40 DIA segments of 30,000 resolution with IT set to 55ms, AGC of 1×106, and a survey scan of 120,000 resolution with 50 ms max IT and AGC of 5×105. The mass range was set to 350-1650 m/z.…”