Container name: vorontsovie/pwmbench_pbm
Benchmark of PBM data developed by Jan Grau. Based on jstacs project.
How to run benchmark:
docker run --rm --mount type=bind,src=$(pwd),dst=/data vorontsovie/pwmbench_pbm <METRIC> /data/pbm_data.txt /data/motif.mat
METRIC corresponds to different metrics.
- ASIS - correlation of intensity (taken as is) and log-sum-occupancy.
- EXP - correlation of intensity and sum-occupancy.
- LOG - correlation of log-intensity and log-sum-occupancy.
- ROC - AUC ROC
- PR - AUC PR
- ROCLOG - AUC ROC in logarithmed intensities
- PRLOG - AUC PR in logarithmed intensities
- all - calculate every metrics ({metrics_name: value, ...} in JSON format is printed)
Motif score is defined as sum-occupancy.
Motif have to be positional frequency matrix or positional count matrix. It's internally corrected with pseudocount of 0.0001.
Motif format (motif.mat
):
>AHR_HUMAN.H11MO.0.B
0.26314477735061276 0.11859745819448689 0.36641340309888065 0.25184436135601973
0.07065186803239253 0.07710428433167413 0.2251456131484061 0.6270982344875272
0.14105450572642986 0.2850314719772257 0.13449544731512686 0.4394185749812176
0.0165402342545919 0.008555293579961341 0.9474208905975482 0.027483581567898402
0.0 0.9766155308814389 0.009695999390622859 0.013688469727938113
0.022350405419921632 0.005133176147976808 0.9702350068107785 0.0022814116213230235
0.0 0.02235040541992163 0.004562823242646056 0.9730867713374324
0.0 0.0 1.0 0.0
0.279810224889837 0.43437346977265195 0.10496230952256 0.180853995814951
PBM data format (pbm_data.txt
):
370241.822966 GAACAATGTAAATTATTGAAAGGGCTAATTCAATTAGTCTGTGTTCCGTTGTCCGTGCTG
365545.496066 GATAACCGACGCCCATTAATTATATTAGCATTGAGCGTCTGTGTTCCGTTGTCCGTGCTG
353259.608349 ATTGATTGATGGCTAACTAAATTAAGCGCATGGAGGGTCTGTGTTCCGTTGTCCGTGCTG
347894.963074 CGTCTATTTTCGGGTAATTATCTCATAATGAGGTGGGTCTGTGTTCCGTTGTCCGTGCTG
334081.174145 GATGCTCCGGATTATTAAGTAATTAAATGAGTTTCCGTCTGTGTTCCGTTGTCCGTGCTG
...