GWAVA looks like a good tool to the prediction of the functional impact of non-coding genetic variants (http://www.sanger.ac.uk/science/tools/gwava). But in In the documentation it only mentions support for GRCh37.
Does anyone know if there is support for GRCh38? (I tried to send an email to the GWAVA team but the email provided isn't working)
Most of those tools for predicting the deleteriousness / functionality of non-coding variants were built on hg19 / GRCh37 data, unfortunately. I am not sure of their support for hg38.
You may want to look through the other tools to see if they support it:
CADD
DANN
FATHMM-MKL
SuRFR
PRVCS
Genomiser
LINSIGHT
Also, in my own opinion, GWAVA actually underperforms on real World data when compared to most of the other tools, but this is because it was mostly trained on TSS variants. Each of these tools targets a particular 'niche', except for CADD, which has broad usage and provides a general 'feel' of what may or may not be functional.