Version 17.1 and higher
Adding de novo transcripts of a known gene to the tranSMART dictionary works basically in the same way as for regular transcripts.
Regular transcripts are loaded with two dictionary files. Both contain Ensembl transcript IDs, one dictionary linking it to associated transcript names and the other that links it to an Entrez gene ID. Since de novo transcripts will neither have an official ID nor an associated name, these can be chosen by the uploader (they do have to be unique).
Adding de novo transcripts will not require any reloading of previously uploaded transcripts, simply providing the new ones will suffice.
For example, if you would like to add two de novo transcripts that belong to TP53 (Entrez gene id: 7157) you could do so by creating the following tsv files.
Then, from the transmart-loader directory run in this order:
java -cp target/loader-jar-with-dependencies.jar org.transmartproject.pipeline.dictionary.TranscriptDictionary <path_to_file>/transcript_id_name_mapping.txt
java -cp target/loader-jar-with-dependencies.jar org.transmartproject.pipeline.dictionary.GeneTransDictionary <path_to_file>/transcript_id_gene_mapping.txt