Output Column Descriptions

Below are descriptions of all column added by DART-ID onto your input data.

Note, if you are providing peptide-level data (e.g., MaxQuant’s evidence.txt instead of msms.txt), then all values are for the best peptide and not all its PSMs.

Confidence Update

dart_PEP	The updated posterior error probability (PEP) for a peptide after applying DART-ID. If this peptide did not participate in DART-ID (due to not enough observations, or decoy tag), then this value defaults to its original PEP as assigned by the search engine.
dart_qval	The q-value (false discovery rate, FDR) derived from the dart_PEP values. To filter at 1% FDR, for example, throw out all peptides with dart_qval > 0.01.

Protein Inference:

The Fido protein inference algorithm stores protein IDs and their associated probabilities in the protein_fdr.txt file in the output folder. In addition, we provide each peptide with the probability of its razor protein:

razor_protein_fdr

The probability that the peptides razor protein (as assigned by the search engine) is present in the sample.

Diagnostic Columns

pep_new	The same as dart_PEP, except that for peptides that did not participate in DART-ID, this value is NA.
participated	True/False, whether or not this peptide participated in the DART-ID algorithm.
exclude	True/False, whether or not this peptide was excluded from alignment but still allowed to participate in the confidence update process (i.e., if True, this peptide was not used to infer retention times, but still had its error probability updated using the inferred RTs from the model)
mu	The reference RT for this peptide.
muij	The inferred retention time for this peptide i in experiment j, derived from mu and the experiment transforms beta_0, beta_1, beta_2, and split_point.
sigmaij	The standard deviation (sqrt(variance)) of RT for this peptide i in experiment j. Derived from mu and the experimental transforms, sigma_slope and sigma_intercept.
rt_minus	The P(RT | ID is incorrect) portion of Bayes’ Theorem. This is derived from evaluating the observed RT on the null distribution of RTs.
rt_plus	The P(RT | ID is correct) portion of Bayes’ Theorem. This is derived from evaluating the observed RT on the inferred RT distribution (built from muij and sigmaij).
residual	The difference between the observed RT and the inferred RT (muij).
input_id	ID number associated with the input data file (actual file, not raw file name).
exp_id	ID number associated with the raw file name.
peptide_id	ID number associated with the peptide identity.
stan_peptide_id	Re-mapped peptide_id that is fed into the optimization program STAN.