Prior to Scaffold DDA 7.0, there was an option to load data in Prefiltered mode. This was simply a way of using the search engine's probabilities in Scaffold DDA. This option was available for Proteome Discoverer, MaxQuant, and FragPipe data.
In Scaffold DDA 7.0, we've introduced and released a new Analysis Mode section in the Load Data dialog:
- Standard – Scaffold DDA will run bundled Percolator, use perfect shared evidence grouping, and apply Scaffold DDA FDR analysis thresholding. The Analysis tab remains available for threshold settings, but protein clustering customization is locked in this mode.
- Pass-through – Scaffold DDA will match supported search engine settings for probabilities, grouping, and thresholding. Data not passing those thresholds will be discarded during loading. In this mode the Analysis tab is disabled because the search engine settings are being passed through directly. Pass-through is only available when all queued files support search engine probabilities, grouping, and thresholding.
- Custom – Choose the probabilities, grouping, thresholding, and condensing behavior individually. If Condensing is turned on, thresholding is forced to From Search Engine. Use this mode when you want to mix Scaffold DDA processing with search engine supplied settings.
| Standard setting | Pass-through setting | |
| Probabilities | Run Percolator: After initial loading, a percolator input file will be created for each sample and percolator run to acquire probabilities within that sample. | From Search Engine: Read PSM probabilities from the input results when available, otherwise don’t acquire probabilities. Currently supported for Scaffold 5/Q+S mzIdentML, MaxQuant, Proteome Discoverer, FragPipe, and Sage. |
| Grouping | Perfect Shared Evidence: Form protein groups out of proteins that have the exact same peptide identifications. Discard subset proteins. | From Search Engine: Respect search-engine designated protein groups as closely as possible. Scaffold DDA will form experiment-wide consensus protein groups if each sample has its own grouping. Currently supported for Mascot, Proteome Discoverer, MaxQuant, mzIdentML, and SFDB. |
| Thresholding | FDR Analysis: Scaffold DDA will use the primary scores to hit user-specified FDRs for PSMs and proteins. | From Search Engine: Mark PSMs and proteins as valid if and only if they pass search-engine specified thresholds. Currently supported for Mascot, Proteome Discover, mzIdentML, and SFDB. |
| Condensing | Off: All data is retained, regardless of validity | On: Data that is invalid is not stored in the SFDB. Can greatly reduce file size. If condensing is on, FDR Analysis is no longer possible. |
What May Look Different in Scaffold DDA
Pass-through mode is designed to preserve the identification decisions from the original search or analysis program while letting you use Scaffold DDA for organization, quantification, visualization, and statistical testing. In Scaffold DDA, the pass-through preset uses search-engine probabilities, search-engine grouping, search-engine thresholding, and condensed loading.
That means Scaffold DDA should agree with the source program on the imported identification evidence that passed the source thresholds. It does not mean every table, count, grouping label, FDR summary, or quantitative value will look identical to the original Proteome Discoverer or Mascot report.
What Pass-Through Mode Is Intended to Match
When a file is loaded in pass-through mode, Scaffold DDA reads the probabilities, grouping information, and threshold decisions that are available from the source file. Data that did not pass the source thresholds may be discarded during condensed loading.
After import, the results are represented in Scaffold DDA's own data model for proteins, peptides, spectra, samples, quant channels, rollups, plots, and statistical tests. This is why the same identification evidence can be organized or summarized differently from the original program.
For Proteome Discoverer result files, Scaffold DDA uses the PD confidence and exclusion information that is available in the result database. For example, PSMs are treated as passing only when PD has not excluded them and the match confidence is high. For protein grouping in PD result files, Scaffold DDA uses PD protein-group links and high-confidence protein-group/protein indicators when those fields are present.
For Mascot result files, Scaffold DDA reads Mascot peptide assignments, Mascot threshold information, and Mascot grouping/family relationships where available. The imported evidence is then placed into Scaffold DDA's grouping model.
Protein Grouping Can Look Different
Different programs use different rules for deciding which proteins belong together, which protein is shown first, and how shared peptides are displayed. Scaffold DDA preserves source grouping information where the importer supports it, but it still displays proteins through Scaffold DDA's protein assembly and protein family model.
For Proteome Discoverer results, Scaffold DDA uses the PD protein group identifiers when matching search-engine grouping. If a protein is associated with more than one PD group, Scaffold DDA assigns it deterministically to one group for display. Singleton proteins may also be represented as Scaffold DDA singleton groups. Because of that, PD group order, representative proteins, and total displayed group counts may not match the PD report exactly.
Mascot differences are often more noticeable. Mascot reports protein hits, same-set proteins, subset/similar proteins, and broader protein families. In Scaffold DDA, Mascot same-set proteins are used as protein groups, while broader Mascot family relationships are stored as protein families. As a result, one Mascot family can appear as multiple Scaffold DDA protein groups, and a Mascot hit or family may not map one-to-one to a single Scaffold DDA group.
For example, you may see:
- A Mascot family split across several Scaffold DDA protein groups.
- A different representative, lead, or top-displayed accession.
- Shared peptides assigned, counted, or displayed differently.
- Subset, similar, same-set, or equivalent proteins labeled differently.
- Protein group counts that differ from the original PD or Mascot report.
When comparing Mascot to Scaffold DDA, it is usually more informative to compare the supporting peptide and spectrum evidence than to expect Mascot hit numbers, family numbers, or displayed representative proteins to match exactly.
In certain cases, even if a file was loaded in pass-through mode, it may be useful to switch the grouping algorithm to Scaffold DDA's perfect shared evidence grouping. (Via the menu option Experiment > Protein Clustering > Select protein grouping method...)
FDR Values May Use Different Units
False discovery rate (FDR) values can look different because the original program and Scaffold DDA may be reporting FDR at different levels.
Proteome Discoverer reports PSM-level confidence/FDR information. Scaffold DDA imports that information for pass-through thresholding, but the Scaffold DDA Dashboard's peptide FDR summary is spectrum-level: it counts target spectra and decoy spectra. Internally, Scaffold DDA collapses PSM evidence to one target/decoy classification per spectrum for this Dashboard value.
That distinction matters. A single spectrum can have more than one candidate PSM. PD may report FDR over PSMs, while the Scaffold DDA Dashboard reports over spectra. Those denominators are not the same, so the displayed FDR values and target/decoy counts can differ even when the same identifications were imported. When comparing FDR values, first check which level is being shown. PD PSM-level FDR should not be expected to exactly match Scaffold DDA's Dashboard spectrum-level peptide FDR.
Quantitation Is Expected to Differ
Quantitative values are expected to differ between the original program and Scaffold DDA.
This is intentional. The point of pass-through mode is to preserve identification decisions while allowing Scaffold DDA to apply its own strengths for quantitative analysis. Scaffold DDA may use different rules for organizing samples, selecting quantitative evidence, handling missing values, normalizing measurements, rolling peptide values up to proteins, summarizing replicate groups, and applying statistical tests.
As a result, quantitative tables, fold changes, normalized intensities, channel summaries, and statistical results may not match the original program exactly. Those differences are expected and are part of why the file is brought into Scaffold DDA: to use Scaffold DDA's organization, quantification, visualization, and statistical testing on the imported identifications.
How to Compare Results
When validating a pass-through import, compare results at the right level:
- For identification agreement, compare the imported peptide and spectrum evidence that passed the source thresholds.
- For protein-level differences, remember that grouping, families, singletons, and representative protein selection can vary by program.
- For PD FDR differences, compare PD's PSM-level values to PSM-level evidence, not to the Scaffold DDA Dashboard's spectrum-level peptide FDR.
- For Mascot grouping differences, compare peptides and spectra first, then interpret Scaffold DDA protein groups as same-set groups and Scaffold DDA protein families as broader Mascot family relationships.
- For quantitation, expect Scaffold DDA results to reflect Scaffold DDA's quantitative workflow rather than the source program's quantitative workflow.
In short, pass-through mode is meant to keep the identification decisions from the original program while presenting and analyzing the data through Scaffold DDA. Exact visual, grouping, FDR, and quantitative matches to the original program should not be expected in every view.