For the first htseq-count job still queued (datasets 37 & 38), the input Tophat job needs to have the metadata "auto-detected". Use the link in the yellow box within input dataset 20 to do this. The job should then properly queue and begin to run. If it doesn't start to execute (yellow status color) after 12 hours or so, use the re-run button to launch it again, then permanently delete the original stalled job/datasets to kill them. Rerunning this way (rerun/delete) of grey queued jobs is generally not recommended or even needed for several reasons, but in your case specifically, this might be the only path forward.
For the second htseq-count job still queued (datasets 47 & 48), the input dataset 47 has been deleted/purged, making it unavailable to the tool. You will need to set up another job for this one, with the inputs left as active until the job completes.
Hope this helps, Jen, Galaxy team