Closed zertan closed 3 years ago
Did I understand correctly that you have run the estimateVBExpression from BitSeq to get the abundance estimates and would now like to use the model to get the probabilities for each read mapping to each reference?
If so, estimateVBExpression already computes these probabilities, but it just does not write them. If this is needed, we can of course add an option to write the probabilities.
Antti
Yes this is exactly what I am interested in! I may fork and look into it and send you a pull request. May I also ask about the official publication of this. Is there one planned?
The BIB paper is available at: http://mgen.microbiologyresearch.org/content/journal/mgen/10.1099/mgen.0.000075 (Added the link to README too - thanks!)
Please feel free to fork BitSeq and send a pull request. The code is a bit messy because some variables are reused for multiple purposes to optimise memory usage. The relevant variable should be 'phi' in VariationalBayes.cpp.
Antti
Now implemented in BitSeq for quite a while.
I wonder if it is possible to use parseAlignment --uniform, then use BitSeqVB to estimate relative "transcript" abundance and use this as new input to parseAlignment -e "Expression file" to get probabilities of each read/fragment mapping to each transcript? I also wonder what file format should the "expression file" be in?