search

xushiabbvie / TDtool

6 stars 0 forks source link

Not fully able to replicate TCGA_essentiality_scores.rds output #4

Closed aambioinfo closed 2 months ago

aambioinfo commented 2 months ago

Hello again, I have run both the run_expression_alignment_TCGADEPMAP.R and build_model_predict_TCGADEPMAP.R scripts, but the values in the TCGA_essentiality_scores.rds that I have obtained are a bit different to the ones published in https://figshare.com/articles/dataset/TCGADEPMAP_gene_essentiality_scores/21382011?file=41467224. Could you please let me know if besides these 2 scripts you run any other that might explain the differences in the output? And which version of CCLE_expression_full.csv did you use? Could you confirm that the Achilles_gene_effect.csv version used if from 21Q1? Many thanks!

xushiabbvie commented 2 months ago

Thank you for trying our code. I think it is normal to see some differences across different runs. The reason is there are a few steps using random seeds for initialization. For example, the cross-validation training can be slightly different if you run twice. For this project, we are more identifying the trend instead of actual values. Hope this helps.