From Peyton:
"It looks like I got everything set up for the model. I plugged in the data you provided and just worked with DNA repair. Unfortunately, I think the inputs need to be a little more advanced, as the average gene expression number is quite low and they only differ by minimal amounts throughout the lifespan of the cell. In essence, it's trying to predict the age, say 7.8 hours from 0.08 as opposed to 72 hours from the input 0.079. I think perhaps we need to discuss a better method for inputting the data, like using all the genes instead of one number. Another possibility would be to not normalize but change the way each gene is used. They are all treated equally but if we can place some importance on specific genes it could contribute to the expression score more than others."
Qin response:
"These are good observations. It seems reasonable that the majority of genes are not very informative on lifespan prediction. There are a couple of ways that we might go from here. Dr. Guo did some analysis on yeast transcriptome and proteome during aging, and his results seem to suggest hub genes and essential genes are important. We could also select genes whose deletion leads to substantial changes in lifespan. "
No comments:
Post a Comment