Submodular-Selection-of-Assays (SSA)

Please see the following reference for more details: Kai Wei * , Maxwell W. Libbrecht * , Jeffrey A Bilmes, William S. Noble. "Evaluation and selection of panels of genomics assays." Submitted.

Get the most recent version on github.

• Author: Kai Wei, Maxwell Libbrecht
• Email: kaiwei@uw.edu, maxwl@cs.washington.edu
• Melodi Lab, University of Washington, Seattle

greedy_selection_facility_location.py

Arguments

• List of assay names (--names): A text file with N lines, with each line being an assay name.
  An example of the assay name file can be found in the file assay_names_<type>.txt where the type is one of "histonemods", "tfs" (transcription factors) or "all".

• Similarity matrix (--sim): A text file containing a symmetric square matrix of size N by N, with each entry being non-negative. The file should have N lines, with each line corresponding to a row of the matrix, and with columns delimited by spaces. An example similarity matrix can be found in the file similarity_matrix_<type>.txt.

• Assay types to select from (--selectFrom) (optional): A subset of the assay types that should be selected from. The file should be a newline-delimited list of names, where each name is included in the --names file. If not specified, all assay types are included.

• Output file name (--output): The path for the file where the selected list of items is to stored.

Sample command line

python greedy_selection_facility_location.py --names=assay_names_all.txt --sim=similarity_matrix_all.txt --output=ordered_assay_list.txt

greedy_selection_facility_location_SSA_past.py

Submodular Assay Selection under past setting (SSA-past)

We provide the source code for running assay selection under the past setting, where the goal is to select from a list of assays already performed on a given cell type. The script is found at SSA_past/greedy_selection_facility_location_SSA_past.py.

Sample command line

cd SSA_past;
python greedy_selection_facility_location_SSA_past.py --cellType=K562 --output=ordered_assay_list.txt

similarity_matrix.py

This script computes a similarity matrix over a number of genomics data sets. The genomics data must be in genomedata format.

Arguments

• List of tracks (--tracks): A text file with one line per track. Each row should have three tab-delimited fields: (1) The path to the genomedata archive that contains this track; (2) The name of the track within the genomedata archive; and (3) The desired name this track should go by in the SSA output.

• Coordinates (--tracks): A bed3 file with the coordinates over which the correlation should be computed. A larger coordinate region results in more accurate correlations, but more time and memory usage.

• Output directory (--outdir): Directory for the output files.

Output

The script will put two files in the output directory: similarity.tab and names.txt. These files can be input into greedy_selection_facility_location.py using the --sim and --names arguments respectively.

Sample command line

python similarity_matrix.py --tracks=example_tracks.txt --coords=example_coords.txt --output=out