Decoy-Free FDR Estimation for Mass-Spectrometry

This a package to estimate FDR in mass-spectrometry searching results from search engines using decoy-free approach. This software will analyze the matching scores output from the search engining.

Install

Make sure python is installed, python official website.

Install the pip, python package management tool,

python -m ensurepip --upgrade

Then, install our package,

pip install decoyfree-msfdr

Usage

Regular MS/MS search FDR estimation

decoyfree-msfdr [-h] [-f INPUT_FILE] [-d INPUT_DIR] [-t INPUT_TYPE] [-m EVAL_MODEL]
                [--score-field SCORE_FIELD] [--threads THREADS] [-c CONSTRAINTS] [-s MODEL_SAMPLES]
                [-r RANDOM_SIZE] [--tolerance TOLERANCE] [--show_plotting] [--out_dir OUT_DIR]

Cross-linked peptide MS/MS search FDR estimation

decoyfree-xlmsfdr [-h] [-f INPUT_FILE] [-d INPUT_DIR] [-t INPUT_TYPE] [-m EVAL_MODEL]
                  [--score-field SCORE_FIELD] [--threads THREADS] [-c CONSTRAINTS] [-s MODEL_SAMPLES]
                  [-r RANDOM_SIZE] [--tolerance TOLERANCE] [--show_plotting] [--out_dir OUT_DIR]

options:
  -h, --help            show this help message and exit
  -f INPUT_FILE, --input-file INPUT_FILE
  -d INPUT_DIR, --input-dir INPUT_DIR
  -t INPUT_TYPE, --input-type INPUT_TYPE
                        format of input files, supported formats are csv,tsv,idXML
  -m EVAL_MODEL, --eval-model EVAL_MODEL
                        existing model to be evaluated
  --score-field SCORE_FIELD
                        the field name holding PSM scores
  --threads THREADS     number of threads
  -c CONSTRAINTS, --constraints CONSTRAINTS
                        Choices of constraints to be used
  -s MODEL_SAMPLES, --model_samples MODEL_SAMPLES
                        Number of samples/top scores to be used in modeling
  -r RANDOM_SIZE, --random_size RANDOM_SIZE
                        Number of random starts per skewness setting
  --tolerance TOLERANCE
                        Threshold of the change of the point-wise log-likelihood for the EM algorithm to determine the
                        convergence
  --show_plotting       Show plotting while fitting the model
  --out_dir OUT_DIR     The place to save results

Options

Option	Argument	Default	Description
-f, --input-file	Path	N/A	Path to the search result file
-d, --input-dir	Path	N/A	Path to the directory holding search result files
-t, --input-type	csv, tsv, idXML	idXML	Search result format
-m, --eval-model	Path	N/A	Path to an existing model to be evaluated
-c, --constraints	no_constraint, unweighted_pdf_mode	unweighted_pdf_mode	Choices of constraints to be used
-s	1, 2	2	Number of samples/top scores to be used in modeling
--threads	Integer	1	Number of threads
-r, --random_starts	Integer	2	Number of random starts per skewness setting
--tolerance	Float	1e-8	Threshold of the change of the point-wise log-likelihood for the EM algorithm to determine the convergence
--show_plotting	Bool	False	Show plotting while fitting the model
--out_dir	Path	./results	The place to save results

Examples

Regular MS/MS search with MSGF+ engine

Suppose the MS/MS search result with MSGF+ software is saved in .tsv format, data/sample.tsv. You can run the FDR estimation algorithm with the following command,

decoyfree-msfdr -f data/sample.tsv -t tsv --out_dir results/sample --threads 10

Multiple search results

If you have multiple search results from MSGF+ saved in the 'data/sample/' directory, and you want to use them all together to build a single model, do the following,

decoyfree-msfdr -d data/sample/ -t tsv --out_dir results/sample --threads 10

Note: this will search the directory for all the files with .tsv extension. If you specified other formats, it will search for the files with the corresponding extension.

MS/MS search results with other engines

If you are using another search engine, please specify the following information,

Option	Default	Description
--score_field	EValue	The score field used to model the data
--log_scale	True	Whether to model on the log scale of the data
--neg_score	True	Whether to take negative of the score. In our model, higher score means better. On log-scale, this is done after taking log
--spec_ref_fields	"#SpecFile,SpecID"	Comma separated fields to identify a spectrum uniquely

XL-MS/MS search with OpenPepXLLF engine

Suppose the MS/MS search result with MSGF+ software is saved in .idXML format, data/sample.idXML. You can run the FDR estimation algorithm with the following command,

decoyfree-xlmsfdr -f data/sample.idXML -t idXML --out_dir results/sample --threads 10

Note: Currently, idXML is the only format we support. Please let us know if you need to use another format in issues. I'll add support to that.

XL-MS/MS search results with other engines

If you are using another search engine, please specify the following information,

Option	Default	Description
--score_field	"OpenPepXL:score"	The score field used to model the data
--log_scale	False	Whether to model on the log scale of the data
--neg_score	False	Whether to take negative of the score, on log-scale, this is done after taking log