You might try to extract those info from ftp://mirbase.org/pub/mirbase/CURRENT/genomes/mmu.gff3 and be aware of the strand info of a precursor to make sure you have right coordinates for TSS (specifically, the start position of precursors).

I think in mirbase they provide the pre-miRNA. And I think you are looking for pri-miRNA. You can find TSS of pri-miRNA in PROmiRNA database.

@Tanvir Alam: Actually, you can find the (canonical) start coordinates of precursor miRNA and its corresponding mature miRNA in the gff3 file provided by miRBase by using the identification of miRNA_primary_transcript and miRNA in the 2nd column of the gff3 file. But, of course, if some one can provide CAGE data, then that will be very valuable data since as I knew, the start/end coordinate of miRNA genes are vary within few base pairs.

Thanks for the discussion.

Tanvir was right. The miRNA primary transcript in the miRBase are around 70-120bp long, implying they are pre-miRNAs and the mature miRNAs are around 20bp.

I came across DIANA tools which gives predicted promoters for miRNAs and they are kilo base pairs upstream. I also got access to PROmiRNA database and it looks like it.

have you checked this: http://www.nature.com/ncomms/2014/141210/ncomms6700/abs/ncomms6700.html#affil-auth

Also Mendells lab has produced a useful dataset:

http://genome.cshlp.org/content/25/9/1401.abstract

NOTE: pri-microRNA transcripts are VERY large in some case (KB) and so miRBase won't help you.

In our most recent paper we have produced a new annotation of vertebrate miRNAs and currently we show +13 and +11nt for each pre-miRNAs

http://mirgenedb.org/

Article A Uniform System for the Annotation of Vertebrate microRNA G...

Also here: 

http://m.nar.oxfordjournals.org/content/early/2015/11/18/nar.gkv1254.abstract

Hello Bastain,

This is very helpful. Thanks a lot.