Potential bug in SPiP_functions.r

[Krakada]

Hi @raphaelleman ,

I'm under the impression that these 2 lines should be removed:

SPiP/RefFiles/SPiP_libs/SPiP_functions.r

Lines 1167 to 1168 in cae95fe

	probaSSPhysioMut <- tmpTableSeqPhy$proba[1]
	classProbaSSPhysioMut <- tmpTableSeqPhy$classProba[1]

as they are trying to fill probaSSPhysioMut and classProbaSSPhysioMut from tmpTableSeqPhy instead of tmpTableSeqPhyMut.

probaSSPhysioMut and classProbaSSPhysioMut seem to be already properly set slightly below:

SPiP/RefFiles/SPiP_libs/SPiP_functions.r

Lines 1176 to 1178 in cae95fe

	if(nrow(tmpTableSeqPhyMut)>0){
	probaSSPhysioMut <- tmpTableSeqPhyMut$proba[1]
	classProbaSSPhysioMut <- tmpTableSeqPhyMut$classProba[1]

but that will overwrite them only when nrow(tmpTableSeqPhyMut)>0, if it's not the case, the 2 (wrong?) values set l.1167 to 1168 will remain and have the exact same value as probaSSPhysio and classProbaSSPhysio.

Could you confirm if it's indeed a bug? I could open a Pull request to remove these 2 lines if it helps.

Regards.

[raphaelleman]

Thanks for proofreading the script. I have chosen to systematically infer the value of the physiological splice site of the mutated sequence from the initial score of the wild-type sequence. If the splice site is modified in the mutated sequence, I update the score accordingly. I prefer not to modify the score as the random forest model used by SPiP could be unexpectedly impacted.