0

Would this be a problem I build a gene signature and tested it in my test, train and validation sets. I do observe significant difference in terms of survival when I had take those genes as predictors but which I obtained using lasso regression. Now when I see the difference in term of my clinical groups which in my case are the subtypes I see difference in terms of few groups in my validation where as for my data sets where I had got the signatures there is fair amount of differences

Now my question is would that be problematic, i.e there is differences in survival based on those genes but not much differences in terms of expression difference between clinical groups

Any suggestion or help would be really appreciated.

PesKchan
  • 155
  • 5
  • 1
    There isn't enough detail in the question for you to get a useful answer. Please show results of the models and point out what troubles you about them. Use the code {} tool to format the software output. If you need to keep the specific genes, clinical groups, etc. secret, change the names to generic things like "geneA", "genesetW", "subtypeC" and so on. – EdM Apr 17 '23 at 13:43
  • @EdM I kind of sorted it out now this is the question I would require your suggestion and help if any https://stats.stackexchange.com/questions/613879/how-to-compare-two-signature-for-prediction-models – PesKchan Apr 24 '23 at 08:46

0 Answers0