Segmentation fault (core dumped)

[AMA-cs]

Hello,

I'm running IntaRNA with options similar to RNAup while integrating shape data. I noticed that when I change the tShapeMethod from "D" to "Z" and the tShapeConversion from "O" to "M", I get a segfault error.

This command works fine:
IntaRNA -v --energyVRNA=/home/ubuntu/ViennaRNA-2.5.1/misc/rna_turner2004.par --model=P --mode=M --noSeed --accW=0 --accL=0 --intLenMax=25 -t targetInput.fasta -q queryInput.fasta --tShape=targ.shape --tShapeMethod="D" --tShapeConversion="O" --outMode=E

However, changing the conversion and shape method results in an error:
IntaRNA -v --energyVRNA=/home/ubuntu/ViennaRNA-2.5.1/misc/rna_turner2004.par --model=P --mode=M --noSeed --accW=0 --accL=0 --intLenMax=25 -t targetInput.fasta -q queryInput.fasta --tShape=targ.shape --tShapeMethod="Z" --tShapeConversion="M" --outMode=E

The error below:

Even when I use this option: --outMode=C --outCsvCols '*', it gives the same error. However, when I exclude these columns, it works: Eall1,Zall1,EallTotal,Etotal. The target sequence is > 300nt. Not sure if the issue is a memory issue or something else. It only happens when with shape data with method "Z".

[martin-raden]

Hi @Abeermm
thanks for your bug report.

The SHAPE integration is done via the API of the ViennaRNA package. Eventually, all parameters are forwarded to its respective functions and used to guide the accessibility computation (done in the fillByRNAplfold() calls within the verbose output).
It seems, these run fine, but obviously, something went wrong. I would guess, the Vienna lib returns some insane values, which subsequently crash the predict() call.

@RaumZeit any ideas from the ViennaRNA lib man?

I will have to have a look what in detail goes wrong, but my guess is artificial unpaired probabilities outside of the range [0,1]..

Respective checks are only enabled in debug mode. So if you are curious, you could recompile IntaRNA using --enable-debug and rerun. It will hopefully fail with some more informative error message.

Will try to find some time soon to look it up myself too.

Best,
Martin