search

ren-lab / deltaSVM

Codes to run deltaSVM models to predict allelic TF binding
GNU General Public License v3.0
8 stars 7 forks source link

SVM models missing #2

Closed BeeSting-pgm closed 3 years ago

BeeSting-pgm commented 3 years ago

The complete set of models are not present in the all_models folder

yjqiu commented 3 years ago

Can you comment on which models are missing?

BeeSting-pgm commented 3 years ago

in the all_models folder they don't contain files, just paths.

yjqiu commented 3 years ago

@BeeSting-pgm Thanks for spotting the missing files. The files have been updated now.

yjqiu commented 3 years ago

Looks like the file is too large for our GitHub space. I have updated the link to our server. @BeeSting-pgm @rmd13

rmd13 commented 3 years ago

Hi, if to use all models, does the obs_threshold and pbs_threshold match all transcription factors, or only the 94 ones？ Thanks

yjqiu commented 3 years ago

@rmd13 No, the thresholds only work for 94 confident models. We don't suggest users to use all models since they are not validated or failed during our benchmark. They are only for advanced users to explore instead of making predictions.

rmd13 commented 3 years ago

In the results table there are 2 columes: "deltaSVM_score" and "preferred_allele". Often there are SNP-TF lines shows relative high value but the "preferred_allele" shows "None", compared to the relative lower "deltaSVM_score" but "Gain". How to interpret this scenario? Does the "Gain" means allelel bind stronger to SNP than allele1? But sometims allele2_bind is -0.706563 but allele1_bind is 0.562300, lookes the reverse. Thanks.

yjqiu commented 3 years ago

@rmd13 Do you mean for additional models or 94 models? For additional models, run.sh will not work. I have updated the readme to include more details. For 94 models, do you mind create a new issue since it is a separate problem.

yjqiu commented 3 years ago

Close issues.