It depends on what sequences you have and what you want to do.
Some web interfaces allow you to have an easy access to upstream region of the gene sequences, then you can download the 2000 or 1500 or 1000 bp region upstream your gene of interesrt.
If you are looking on-line tools for promoter prediction, I think that usually people take the region upstream their gene of interest and define an arbitrary length of the promoter (1000 bp to 2000 bp)
Your question evokes the notion of computational prediction of genome function, a capacity that yet eludes science. For instance, there is not now a computational method of predicting the Origin of Replication (ORI). Indeed, the development of software capable of such genomic feature detection is an active research area. Hence, if the foregoing describes well your need, then we may reasonably suggest that you will not find such software.
As David Cohen pointed out around 2 kb upstream of the start of the ORF is generally considered or initially assumed as the promoter unless experimentally proven otherwise. How you can get that sequence is: If you know the gene coordinates (or annotated gene coordinates if the gene is not characterized) of your gene of interest in a particular chromosome then get the sequences 2 Kb upstream of the start.
The 2KB upstream strategy is one of the "assumption" based process and speaking from experience, it really gives assumed result. Initially when I tried this, looking from the DNA sequence of target gene from ENSMBL, I found it really difficult. Many a times intronic regions gets amplified and stuffs like that.
I agree with Somnath but when you do not have much characterization info that is a safe way to start. It paid off many a times in our hands and others I know. One caveat could be having another gene with in this region (depends on your genome such as yeast genomes). You need to be careful about that.
go to ensembl website: http://www.ensembl.org/index.html
choose an organism such as human http://www.ensembl.o…iens/Info/Index
Search your gene such as BRCA2 http://www.ensembl.o…ns;idx=;q=brca2
Click the right hit on the search result page and it will bring you to the gene summary page. For example the link to BRCA2 gene is http://www.ensembl.o…ns;idx=;q=brca2
On the left, under “Gene Summary”, click “Sequence”, the sequence of the gene including 5′ flanking, exons, introns and flanking region will be displayed.
The exons are high lighted in pink background and red text, the sequence in front of the first exon is the promoter sequence.
By default, 600 bp 5′-flanking sequence (promoter) is displayed. If you want to get more, click “Configure this page” in the lower left column, a popup window opens allowing to input the size of 5′ Flanking sequence (upstream). You can put for example “1000” and then save the configuration.
Sometimes there are discrepancies between Ensembl and UCSC annotation regarding TSS. To make sure the first exon given by ensembl is right, copy the promoter sequence
Go to UCSC BLAT search at http://genome.ucsc.e…t?command=start and choose the right genome (eg, human), paste the sequence there. On the result page, click browse of the first hit, this will bring you to the genome browser Page. the query sequence is now aligned with UCSC genome sequence. Zoom out a bit, you will be able to determine whether the promoter sequence matches UCSC annotation. If it matches, the sequence is very likely the right one. Here is the BRCA2 promoter sequence aligned to BRAC2 gene.
In UCSC genome broswer, you can turn on CpG island feature, if there is CpG island in the promoter sequence, the sequence is highly likely a true promoter. In the above example (BRCA2), a CpG island is displayed in the proximal promoter.
Beware some genes have alternative promoters. To find those sequences, it requires extensive bioinformatics and experimental analysis