If you use TCGA data you may want to check out my new TCGA data portal: http://www.oncolnc.org/
I have a preprint describing OncoLnc here: https://peerj.com/preprints/1780v1/
I also have a video showing how to use OncoLnc:
And I have some information at my website: http://www.omnesres.com/research/oncolnc/
OncoLnc focuses on survival correlations with expression data, and the expression data is only for tumors (no normals), and is always RNA-SEQ except for the GBM miRNA data for which there was only microarray data available.
OncoLnc contains precomputed Cox regressions, which were in general computed with an equation like this: coxph(Surv(times,died) ~ gene + lowgrade + highgrade + sex + age. Sometimes cancers did not have grade information, so not every cancer will have grade in the model.
All the code for performing the Cox regressions is available at my GitHub: https://github.com/OmnesRes/onco_lnc, and every Cox regression performed, including ones not in the OncoLnc database, are available as supplemental information in the publication.
OncoLnc will give you the results of the Cox analysis for up to 21 cancers for your gene of interest, which can be either a Tier 3 TCGA mRNA, a mature miRNA, or a MiTranscriptome lncRNA.
OncoLnc also allows for the generation of publication quality Kaplan-Meier plots and downloading of the data used in the plots.
I hope you guys find the site useful!