Yes, it's possible, but it depends on what you know or are willing to learn. I'm sure there are a variety of ways to do it. One method using R would be to use the GRanges library to manage all your features - for instance if you have gene definitions in gff, or at biomart, you can read them into GRanges and then define TSS by resizing all start coordinates to be the size you consider a TSS to be, and then use those GRanges to extract coverage from either bigWig files, or a coverage object, of your RNA Seq data. There are a variety of ways to do this step as well, some of them are more confusing than others. You would tweak these steps for your various feature types. Once you build a matrix of coverages across your uniform features, the metagene is simply the column averages.

If you like R, look up how to use GRanges, GenomicFeatures, and rtracklayer.