This paper is primarily for #CRISPR researchers rather than computational biologists expert in #MachineLearning because well, there is a lot going on in this space and it is very easy to be lost with the terminology and ML algorithms if you are not familiar with ML
-
-
Prikaži ovu nit
-
A crucial aspect we discussed is data labelling. What is the size of the training dataset? how to define the features for the training model? how to label it & avoid the major pitfall, which is the unbalanced training dataset. Well, the details are in the paper.
pic.twitter.com/hhHbZoqCzu
Prikaži ovu nit -
Another aspect we discuss is the features to include into
#MachineLearning Algorithm. Do we use only guideRNA sequence, nucleotide composition at target site, epigenetic information... How many features to define? Well largely depends on your training dataset.Prikaži ovu nit -
Next is how to translate your features from a
#CRISPR experiment into a#MachineLearning readable features. Well this involves tokenizationpic.twitter.com/i0j29MbF43
Prikaži ovu nit -
And of course, a very important aspect is the choice of the
#MachineLearning algorithm for a#CRISPR experiment. Well it depends a lot on the type of experiment, the size of the training dataset and so on.pic.twitter.com/UCMsymzmUz
Prikaži ovu nit -
We finally made series of recommendations on the use of
#MachineLearning for#CRISPR experiments going forward. Please feel free to read our paper. It is in#openaccess and please feel free to download it.Prikaži ovu nit
Kraj razgovora
Novi razgovor -
-
-
Congratulations everyone


-
Thanks Anastasia !
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.