If you are using the Tuxedo suite of tools, then iGenomes is one good
resource. The NCBI files represent RefSeq, and RefSeq includes
NM_XXXX = coding transcript with defined p_id, tss_id, gene_name
NR_XXXX = non-coding transcript with defined tss_id, gene_name
The mm9 version of the iGenomes GTF file is already on the public Main
Galaxy server at:
Shared Data -> Data Libraries -> iGenomes
More about the source is here, and is also where you can obtain mm10
(download tar locally to your computer, then just upload the genes.gtf
file as a dataset):
If using other software, you have many other good choices. Most will
have user groups that suggested the best fit for the tool. Reviewing
'Gene and Gene Prediction' tracks at UCSC (http://genome.ucsc.edu)
give you an idea of those that contain non-coding, predicted, etc.
transcripts. Biomart is also a great resource where Ensembl and other
annotation in GTF or BED format can be reviewed and retrieved.