"ValueError: could not assign tuple of length 1 to structure with 16 fields." when using -b option

[billadelonge]

Hi Ryan,
Thank you for updating this!
I'm having trouble running the blast and keep getting this error "ValueError: could not assign tuple of length 1 to structure with 16 fields."
I believe it has to do with the numpy version and how it works with the other dependencies (maybe?). Could either share an environment for the probemaker or provide the versions of the packages you installed?
Thanks for your help!
Best,
Laura

[billadelonge]

Okay so I think it may have to do with the blast version, I had to manually change the delimiter for the blast cmd to "\t" instead of "$" it works now :)

[billadelonge]

However, it seems that probemaker removes the last probe set(s) and not the ones that have out of target hits. Sorry for the spam I have a complicated order to make and wanted to make sure my probes were not hitting other genes.

[rwnull]

Hey Laura, thanks for reaching out.
I am glad you could resolve the first error. As for the blast function, I have coded it to flag potential off target probe sets, but since there are so many variables when deciding if something is truly off target, I decided to leave the decision to the user. Often when I have used the blast function, the flagged sequences turn out to be isoforms of the same orf, which is not an issue for me but maybe something the user wants to avoid.

As of now I am calling it a feature not a bug, I can see having the option to drop flagged probe pairs as being useful though.

I will leave this issue open to get your input on it.

[billadelonge]

Heya!
Sorry for the double spam, thought it may be worth having it here if anyone has the same issue. I can see the feature being useful actually and totally agree it is best to not remove the probe pair automatically. But I think currently when blast does find a hit it then removes the last probe pair. Let me know if you can replicate this!
Thanks for your help!

[rwnull]

Absolutely!
I poked around and found what was causing the program to omit the final PP, thanks for pointing it out! The repo as it stands should have squashed it.
You can replace CLS.py and mainscript_core.py on your machine with the corresponding files on the repo.

Give it a spin and let me know if it is working for you.

Thanks again for bringing it to my attention!

[desmondramirez]

Hi Ryan!
Finally getting around to using this version of the probe maker. I ran into the same initial error as Laura did. I checked the blast command in the current version of the script and it looks like the delimiter doesn't need to be escaped, so delimiter='\t', rather than '\t'. That allows blastn to work and I get this output:

Probe pairs that had possible off-target matches to the provided database (lower e-value but with high site coverage). 
   Pairs 
[]

Probe pairs that had good matches to the provided database were determined to be the following.
   Pairs 
['1' '10' '11' '12' '13' '14' '15' '16' '17' '18' '19' '2' '20' '21' '22'
 '23' '24' '25' '26' '27' '28' '29' '3' '30' '31' '32' '33' '34' '35' '4'
 '5' '6' '7' '8' '9']



TypeError: slice indices must be integers or None or have an __index__ method

I am bad at python and haven't been able to figure out what's causing this error yet. So I can make the probes if I don't run the blastn search, but we've found some probesets that have weirdly specific, nuclear, off-target labeling, and we've resolved that issue every time we've tried by making a new set of probes with the homology blast.

I'll use an older tool to make my current order, but wanted to give you the heads up on this.