Ideas

• Compare text vs. semantic similarity
• Bake-off various language distance metrics
• Various method of distance too
• Compare to random

Can you use text comparison methods to find similarities between human diseases and zebrafish phenotypes?

Or do you need a very custom mapping via ontologies? The original design was to do something like this. Good idea to first do a bunch of comparisons and then determine if you need to develop something new.

• Download a bunch of python-based text comparision libraries
• Start some simple bake-offs
• Figure out how to do the random model
• Which are our gold standards?

Q: What does failure look like? A: Random is indistinguishable from real diseases

Q: What does success look like? A: Gold standards are found

Compare human diseases vs. human diseases Compare zf phenotypes vs zf phenotypes

textcompare1.py

install conda pip3 install nltk scikit-learn transformers torch fasttext

curl https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.en.300.bin.gz --output cc.en.300.bin

textcompare2.py

pip3 install -U sentence-transformers

works

textcompare3.py

pip3 install tensorflow tensorflow_hub

works