/STRprofiler

A python package, command-line tool, and Shiny application to compare short tandem repeat (STR) profiles.

Primary LanguageHTMLMIT LicenseMIT

STRprofiler

Coverage Status PyPI version PyPI pyversions PyPI license DOI

STRprofiler is a python package, CLI tool, and Shiny application to compare short tandem repeat (STR) profiles. In particular, it is designed to aid research labs in comparing models (e.g. cell lines & xenografts) generated from primary tissue samples to ensure contamination has not occurred. It includes basic checks for sample mixing and contamination and provides a simple interface to conveniently query the Cellosaurus database via the CLASTR API.

STRprofiler is intended only for research purposes.

For each STR profile provided, STRprofiler will generate a sample-specific report that includes the following similarity scores as compared to every other profile:

Tanabe, AKA the Sørenson-Dice coefficient:

$$ score = \frac{2 * no.shared.alleles}{no.query.alleles + no.reference.alleles} $$

Masters (vs. query):

$$ score = \frac{no.shared.alleles}{no.query.alleles} $$

Masters (vs. reference):

$$ score = \frac{no.shared.alleles}{no.reference.alleles} $$

Amelogenin is not included in the score computation by default but can be included by passing the --score_amel flag.

Installation

STRprofiler is available on PyPI and can be installed with pip:

pip install strprofiler

Usage

STRprofiler can be run directly from the command line.

strprofiler compare -sm "SampleMap_exp.csv" -scol "Sample Name" -o ./strprofiler_output STR1.xlsx STR2.csv STR3.txt

Full usage information can be found by running strprofiler --help.

 Usage: strprofiler compare [OPTIONS] INPUT_FILES...   

 STRprofiler compares STR profiles to each other.  

╭─ Options ────────────────────────────────────────────────────────────────────────────────╮
│ --tan_threshold    -tanth   FLOAT    Minimum Tanabe score to report as potential matches |
|                                      in summary table. [default: 80]                     │
│ --mas_q_threshold  -masqth  FLOAT    Minimum Masters (vs. query) score to report as      |
|                                      potential matches in summary table. [default: 80]   │
│ --mas_r_threshold  -masrth  FLOAT    Minimum Masters (vs. reference) score to report as  |
|                                      potential matches in summary table. [default: 80]   │
│ --mix_threshold    -mix     INTEGER  Number of markers with >= 2 alleles allowed before  |
|                                      a sample is flagged for potential mixing.           |
|                                      [default: 3]                                        │
│ --sample_map       -sm      PATH     Path to sample map in csv format for renaming.      |
|                                      First column should be sample names as given in     |
|                                      STR file(s), second should be new names to assign.  | 
|                                      No header.                                          │
│ --database         -db      PATH     Path to an STR database file in csv, xlsx, tsv,     |
|                                      or txt format.                                      │
│ --amel_col         -acol    STR      Name of Amelogenin column in STR file(s).           |
|                                      [default: 'AMEL']                                   │
│ --sample_col       -scol    STR      Name of sample column in STR file(s).               |
|                                      [default: 'Sample']                                 │
│ --marker_col       -mcol    STR      Name of marker column in STR file(s).               |
|                                      Only used if format is 'wide'. [default: 'Marker']  │
│ --penta_fix        -pfix    FLAG     Whether to try to harmonize PentaE/D allele         |
|                                      spelling. [default: True]                           │
│ --score_amel       -amel    FLAG     Use Amelogenin for similarity scoring.              |
|                                      [default: False]                                    │
│ --output_dir       -o       PATH     Path to the output directory.                       |
|                                     [default: ./STRprofiler]                             │
│ --version                            Show the version and exit.                          │
│ --help                               Show this message and exit                          │
╰──────────────────────────────────────────────────────────────────────────────────────────╯

CLASTR

Additionally, the Cellosaurus (Bairoch, 2018) cell line database can be queried via the CLASTR (Robin, Capes-Davis, and Bairoch, 2019) REST API.

strprofiler clastr -sm "SampleMap_exp.csv" -scol "Sample Name" -o ./strprofiler_output STR1.xlsx STR2.csv STR3.txt

Full usage information can be found by running strprofiler clastr --help.

 Usage: strprofiler clastr [OPTIONS] INPUT_FILES...   

**strprofiler clastr** compares STR profiles to the human Cellosaurus knowledge base using the CLASTR REST API.

╭─ Options ────────────────────────────────────────────────────────────────────────────────╮
│ --search_algorithm  -sa    INT  Search algorithm to use in the CLASTR query.             |
|                                 1 - Tanabe, 2 - Masters (vs. query);                     |
|                                 3 - Masters (vs. reference) [default: 1]                 │
│ --scoring_mode      -sm    INT  Search mode to account for missing alleles in query or   |
|                                 reference. 1 - Non-empty markers, 2 - Query markers,     |
|                                 3 - Reference markers. [default: 1]                      │
│ --score_filter      -sf    INT  Minimum score to report as potential matches in          |
|                                 summary table. [default: 80]                             │
│ --max_results       -mr    INT  Filter defining the maximum number of results to be      |
|                                 returned. [default: 200]                                 │
│ --min_markers       -mm    INT  Filter defining the minimum number of markers for        |
|                                 matches to be reported. [default: 8]                     │
│ --sample_col        -scol  STR  Name of sample column in STR file(s).                    |
|                                 [default: 'Sample']                                      │
│ --marker_col        -mcol  STR  Name of marker column in STR file(s).                    |
|                                 Only used if format is 'wide'. [default: 'Marker']       │
│ --penta_fix         -pfix  FLAG Whether to try to harmonize PentaE/D allele spelling.    |
|                                 [default: True]                                          │
│ --score_amel        -amel  FLAG Use Amelogenin for similarity scoring. [default: False]  │
│ --output_dir        -o     PATH Path to the output directory. [default: ./STRprofiler]   │
│ --version                       Show the version and exit.                               │
│ --help                          Show this message and exit.                              │
╰──────────────────────────────────────────────────────────────────────────────────────────╯

Input Files(s)

STRprofiler can take either a single STR file or multiple STR files as input. These files can be csv, tsv, tab-separated text, or xlsx (first sheet used) files. The STR file(s) should be in either 'wide' or 'long' format. The long format expects all columns to map to the markers except for the designated sample name column with each row reflecting a different profile, e.g.:

Sample D1S1656 DYS391 D3S1358 D2S441 D16S539 D5S818 D8S1179 D22S1045 D18S51
Line1 12,14 12 13 12,14 17.3 16,17 18.3 17,11
Line2 12,14 11.3,12 13,15 12,14 17.3 16,17 18.3 17,11
...

The wide format expects a line for each marker for each sample, e.g.:

Sample Name Marker Allele 1 Size 1 Height 1 Allele 2 Size 2 Height 2 Allele 3
Sample1 DYS391
Sample1 D3S1358 16 128.29 8268 18 136.84 5467 16
Sample1 D16S539 12 110.7 9660 13 115.17 5215
Sample1 Penta D 9 415.04 5099 13 435.88 9426
Sample1 D22S1045 15 455.95 13504 17 462.06 6186
Sample1 Penta E 11 397.7 7420 14 412.02 5986
Sample1 D18S51 12 153.72 9134 16 170.48 10501
Sample1 D2S1338 20 263.91 3209 21 267.97 3834
Sample1 TH01 7 85.33 8305 9.3 97.43 7853
Sample1 D7S820 10 292.51 12340 14 308.71 11784
Sample1 D12S391 15 141.53 12870 18.3 157.12 13731
Sample1 AMEL X 81.97 16696
Sample1 D10S1248 16 283.82 8469
Sample1 D13S317 12 328.21 7079
Sample1 D21S11 32.2 239.67 19231
Sample1 TPOX 11 424.02 12239
Sample1 D19S433 14 228.37 14273
Sample1 FGA 23 302.23 14599
Sample2 D16S539 9 97.59 9286 11 106.43 8592
Sample2 TH01 9.3 97.45 5920
Sample2 D8S1179 13 101.1 26414
Sample2 AMEL X 82.1 7476 Y 88.34 8029
Sample2 D3S1358 14 119.87 10146 15 124.14 10160 19
Sample2 D18S51 12 153.8 9316 18 178.79 9182 19
Sample2 Penta D 10 420.13 7693 11 425.25 7945 12
Sample2 vWA 17 156.9 7953 18 160.86 8230
Sample2 TPOX 9 416 6596 11 424.02 5304
Sample2 D12S391 21 166.75 13481 22 170.9 14232
Sample2 D22S1045 15 455.95 14310 17 462.06 10898
Sample2 D2S441 14 236.24 18628
Sample2 DYS391 10 468.83 6722
Sample2 FGA 21 294.67 11941

In this format, the marker_col must be specified. Only columns beginning with "Allele" will be used to parse the alleles for each sample/marker. Any other size or height columns will be ignored.

Output Files

STRprofiler generates two types of output files. The first is a summary file, which contains the top hits for each sample above the specified scoring thresholds. This file provides a useful overview in addition to a flag to identify samples with potential mixing for closer inspection. In the output directory, this file will be named full_summary.strprofiler.YYYYMMDD.HH_MM_SS.csv where the date and time are the time the program was run.

In addition to the marker columns, the summary file contains the following columns:

Column Name Description
mixed Flag to indicate sample mixing.
top_hit Name and Tanabe score of top match to sample.
next_best Name and Tanabe score of next best match to sample.
tanabe_matches Name and Tanabe score of matches above scoring threshold to sample.
masters_query_matches Name and Masters (vs. query) score of matches above scoring threshold to sample.
masters_ref_matches Name and Masters (vs. reference) score of matches above scoring threshold to sample.

The second is a sample-specific comparison file, which contains the results of the comparison between the query sample and all other provided samples. These files are generated for each STR profile provided in the input file(s) and named after the query sample in question. For example, if the input file contains a sample named Sample1, the output file will be named Sample1.strprofiler.YYYYMMDD.HH_MM_SS.csv.

In addition to the marker columns, this output contains the following columns:

Column Name Description
mixed Flag to indicate sample mixing.
query_sample Flag to indicate query sample.
n_shared_markers Number of shared markers between query and reference sample.
n_shared_alleles Number of shared alleles between query and reference sample.
n_query_alleles Total number of alleles in query sample.
n_reference_alleles Total number of alleles in reference sample.
tanabe_score Tanabe similarity score.
masters_query_score Masters (vs query) similarity score.
masters_ref_score Masters (vs reference) similarity score.

clastr

Output for strprofiler clastr is provided in XLSX format. Results follow the CLASTR format, documented here: https://www.cellosaurus.org/str-search/help.html#4

Database Comparison

STRprofiler can be also used to compare batches of samples against a larger database of samples.

strprofiler compare -db ExampleSTR_database.csv -o ./strprofiler_output STR1.xlsx

In this mode, inputs are compared against the database samples only, and not among themselves. Outputs will be as described above for sample input(s).

Database Format

The database should be formatted as a samples by markers matrix and saved as a csv file, e.g:

Sample Amelogenin CSF1PO D13S317 D16S539 D18S51 D19S433 D21S11 D2S1338 D3S1358 D5S818 D7S820 D8S1179 FGA TH01 TPOX vWA PentaE PentaD
sample1 X,Y 12 8 13 14 14 31,31.2 17,19 15 11,12 11,12 12,15 23 7,9.3 8 18
sample2 X 10 9 13 16 12,14 29 20,23 15,16 12,13 9,12 14,15 18 7 8,9 15

Optionally, one may provide two metadata columns - "Center" and "Passage", which will be recognized as non-marker columns.

The STRprofiler App

New in v0.2.0 is strprofiler app, a command that launches a Shiny application that allows for user queries against an uploaded or pre-defined database (provided with the -db parameter) of STR profiles.

This application can provide a convenient portal to a group's STR database and can be hosted on standard Shiny servers, Posit Connect instances, or ShinyApps.io.

An example of the application can be seen here.

Deploying an strprofiler App

Building an app for deployment to any of the above options is simple.

First, make your app.py file:

from strprofiler.shiny_app.shiny_app import create_app

database = "./tester_db.csv"
app = create_app(db=database)

If no database is provided, an example database included with the package will be used. The database file uses same format as for the standard strprofiler command.

Then create a requirements.txt file in the same directory with strprofiler listed:

strprofiler>=0.4.1

This app can then be deployed to any of the above endpoints as one would with any other Shiny app, e.g.:

rsconnect deploy shiny -n your_server -t STRprofiler .

Alternatively, one could export it as a shinylive app and host it on Github pages or similar.

Contributing

You can contribute by creating issues to highlight bugs and make suggestions for additional features. Pull requests are also very welcome.

License

STRprofiler is released on the MIT license. You are free to use, modify, or redistribute it in almost any way, provided you state changes to the code, disclose the source, and use the same license. It is released with zero warranty for any purpose and the authors retain no liability for its use. Read the full license for additional details.

References

If you use STRprofiler in your research, please cite the following:

Jared M Andrews, Michael W Lloyd, Steven B Neuhauser, Margaret Bundy, Emily L Jocoy, Susan D Airhart, Carol J Bult, Yvonne A Evrard, Jeffrey H Chuang, Suzanne Baker, STRprofiler: efficient comparisons of short tandem repeat profiles for biomedical model authentication, Bioinformatics, 2024;, btae713, https://doi.org/10.1093/bioinformatics/btae713

If you use the clastr command or functionality from the Shiny application, please cite the Cellosaurus and CLASTR publications:

Bairoch A. (2018) The Cellosaurus, a cell line knowledge resource. Journal of Biomolecular Techniques. 29:25-38. DOI: 10.7171/jbt.18-2902-002; PMID: 29805321

Robin, T., Capes-Davis, A. & Bairoch, A. (2019) CLASTR: the Cellosaurus STR Similarity Search Tool - A Precious Help for Cell Line Authentication. International Journal of Cancer. PubMed: 31444973  DOI: 10.1002/IJC.32639