Writing about natural science, math, academia and more. https://tavoglc.medium.com/
Don't wanna be here? Send us removal request.
Text
Forecasting COVID19 mutational patterns with variational autoencoders
1 note
·
View note
Text
Want to predict new COVID-19 variants but don't know where to start? Here is how
0 notes
Text
Tired of all those COVID variants? get ahead of the game, and predict them yourself with JAX. #COVID #MachineLearning #Python
https://github.com/TavoGLC/SARSCov2Solar
0 notes
Text
Why is hard to find the seasonality or pattern behind COVID-19 waves? One possibility is the lack of a meaningful coordinate system. #COV1D #Covid
For example, if we look at the movement of mercury through time we will end up observing also the retrograde movement of the planet. That's due to our position relative to the planet. #COV1D #Covid
If we position ourselves in the sun, then we will see the normal movement of all the planets. Luckily neural networks can be used to retrieve a meaningful coordinate system from complex data sources. #COV1D #Covid
In the case of COVID-19, the data source will be the genomic sequences obtained through genomic surveillance. #COV1D #Covid
Training a VAE with a custom sequence representation sports the different genomes by a temporal variable. Currently, the best fit is the sunshine duration rate of change. A proxy measure of the rate of change of solar radiation at a specific location. #COV1D #Covid
If we plot the cases using the sunshine duration as a measure of time the different waves are located at the extremes of the axis. But if we look at how the sunshine duration changes specific rates appear to constrain the duration of the waves #COV1D #Covid
Also by approaching polar latitudes the frequency of those periods increases. #COV1D #Covid
Would you like to know more?
3 notes
·
View notes
Text
Changes in SARSCov2 genome size by date and location.
Would you like to know more ?
0 notes
Text
0 notes
Text
0 notes
Text
0 notes
Text
0 notes
Text
0 notes
Text
0 notes
Text
0 notes