FAQ

Detailed instruction for installing and using Triqler as well as how to interpret and analyze the results can be found in the Triqler user manual

Instruction on general usage of Triqler and an example set can be found on the main page of this repository.

I ran Triqler and received several output files, how do I interpret them?

There is a detailed description of the input and output formats and how to interpret them in the Triqler user manual. For the <input_file>.pqr.tsv and <input_file>.sqr.tsv output files, see the question below.

What are the <input_file>.pqr.tsv and <input_file>.sqr.tsv files and why are the sample columns repeated three times in them?

The <input_file>.pqr.tsv summarizes the results across all samples on peptide level and the <input_file>.sqr.tsv on spectrum level. The three appearances of the sample columns represent:

• 1st set: the match-between-run PEPs, these are all 0 unless you use Quandenser.
• 2nd set: the abundances/intensities, after division by a power of 10 to increase readability (not log transformed!)
• 3rd set: the identification PEPs, calculated from the search engine scores, reported on PSM level.

I have RAW/mzML files, how do I run Triqler on them?

Unfortunately, you cannot run Triqler directly on a RAW or mzML file, or any other spectrum file for that matter. The easiest way to use Triqler would be to process your RAW files with MaxQuant and convert that output to a Triqler input file, as described here. If you prefer another search engine, the Triqler user manual provides an example how to achieve this as well.

I have run a default MaxQuant search with 1% FDR and/or match-between-runs, can I still run Triqler?

Predictions might not be optimal if you used the default 1% FDR cut-offs in your MaxQuant/Andromeda search, since Triqler relies on decoys to estimate identification error rates. We recommend using a 10% or even 100% FDR on all levels (PSM, peptide and protein), but as a first test it should work with 1% FDR as well.

Match-between-runs in MaxLFQ unfortunately does not give error estimates for the matching, which has been shown to be error-prone. However, on several engineered datasets we have tested, the match-between-runs identifications did not result in more false positives.

Can Triqler handle fractionated data?

Support for fractionated data is very rudimentary by simply giving all fractions of the same sample the same name in the run column. Triqler will then only use the best scoring PSM across all fractions. We are open to suggestions for a better scheme for handling fractionated data.

As part of the Triqler converters, the fractions can be specified as part of the --file_list_file file input (see figure 2 in the Triqler manual), which will do the retention time based normalization per fraction and automatically does the grouping by sample as mentioned above.

What is the recommended setting for min_samples? (default = 2)

We recommend setting this to total number of samples / 3. If you set a very low number for min_samples, you could observe more false positives than the reported q-value but this underestimation is usually not dramatic, e.g. 10% observed FDR at 5% reported FDR. We tested this for a human-yeast mixture in the Quandenser paper, see Supplementary Table 2. Note that we used max-missing instead of min_samples in that table, so the first row would be min_samples = 11-8 = 3.

What is the recommended setting for fold_change_eval? (default = 1.0)

The latest version of Triqler reports the minimum advisable fold_change_eval, which is estimated from the data. Typically, this will be somewhere between 0.5 and 1.0. Setting a low fold_change_eval increases the chance for more false positives than the reported FDR, but as long as you stay above the recommended value by Triqler we do not expect major problems. We tested this for several spike-in datasets in the Triqler for MaxQuant paper, see Figure 1 and Supplementary Figure 1.