I wrote a tool named backlocate for this job .

This tool is available in my experimental package http://code.google.com/p/variationtoolkit/.

example:

echo -e  "NOTCH2\tM1T\nEIF4G1\tD240Y" |\
    backlocate -f /path/to/hg19.fa 

#User.Gene    AA1    petide.pos.1    AA2    knownGene.name    knownGene.strand    knownGene.AA    index0.in.rna    codon    base.in.rna    chromosome    index0.in.genomic    exon
##uc001eik.2
NOTCH2    M    1    T    uc001eik.2    -    M    0    ATG    A    chr1    120612019    Exon 1
NOTCH2    M    1    T    uc001eik.2    -    M    1    ATG    T    chr1    120612018    Exon 1
NOTCH2    M    1    T    uc001eik.2    -    M    2    ATG    G    chr1    120612017    Exon 1
##uc001eil.2
NOTCH2    M    1    T    uc001eil.2    -    M    0    ATG    A    chr1    120612019    Exon 1
NOTCH2    M    1    T    uc001eil.2    -    M    1    ATG    T    chr1    120612018    Exon 1
NOTCH2    M    1    T    uc001eil.2    -    M    2    ATG    G    chr1    120612017    Exon 1
##uc001eim.3
NOTCH2    M    1    T    uc001eim.3    -    M    0    ATG    A    chr1    120548116    Exon 2
NOTCH2    M    1    T    uc001eim.3    -    M    1    ATG    T    chr1    120548115    Exon 2
NOTCH2    M    1    T    uc001eim.3    -    M    2    ATG    G    chr1    120548114    Exon 2
##Warning ref aminod acid for uc003fnp.2  [240] is not the same (I/D)
EIF4G1    D    240    Y    uc003fnp.2    +    I    717    ATC    A    chr3    184039089    Exon 10
EIF4G1    D    240    Y    uc003fnp.2    +    I    718    ATC    T    chr3    184039090    Exon 10
EIF4G1    D    240    Y    uc003fnp.2    +    I    719    ATC    C    chr3    184039091    Exon 10
##Warning ref aminod acid for uc003fnu.3  [240] is not the same (I/D)
EIF4G1    D    240    Y    uc003fnu.3    +    I    717    ATC    A    chr3    184039089    Exon 9
EIF4G1    D    240    Y    uc003fnu.3    +    I    718    ATC    T    chr3    184039090    Exon 9
EIF4G1    D    240    Y    uc003fnu.3    +    I    719    ATC    C    chr3    184039091    Exon 9
##Warning ref aminod acid for uc003fnq.2  [240] is not the same (V/D)
EIF4G1    D    240    Y    uc003fnq.2    +    V    717    GTA    G    chr3    184039350    Exon 7
EIF4G1    D    240    Y    uc003fnq.2    +    V    718    GTA    T    chr3    184039351    Exon 7
EIF4G1    D    240    Y    uc003fnq.2    +    V    719    GTA    A    chr3    184039352    Exon 7
##Warning ref aminod acid for uc003fnr.2  [240] is not the same (L/D)
EIF4G1    D    240    Y    uc003fnr.2    +    L    717    CTC    C    chr3    184039581    Exon 6
EIF4G1    D    240    Y    uc003fnr.2    +    L    718    CTC    T    chr3    184039582    Exon 6
EIF4G1    D    240    Y    uc003fnr.2    +    L    719    CTC    C    chr3    184039583    Exon 6
##Warning ref aminod acid for uc003fny.3  [240] is not the same (T/D)
EIF4G1    D    240    Y    uc003fny.3    +    T    717    ACC    A    chr3    184039677    Exon 3
EIF4G1    D    240    Y    uc003fny.3    +    T    718    ACC    C    chr3    184039678    Exon 3
EIF4G1    D    240    Y    uc003fny.3    +    T    719    ACC    C    chr3    184039679    Exon 3
##uc010hxx.2
EIF4G1    D    240    Y    uc010hxx.2    +    D    717    GAT    G    chr3    184038780    Exon 10
EIF4G1    D    240    Y    uc010hxx.2    +    D    718    GAT    A    chr3    184039069    Exon 11
EIF4G1    D    240    Y    uc010hxx.2    +    D    719    GAT    T    chr3    184039070    Exon 11
##Warning ref aminod acid for uc003fns.2  [240] is not the same (L/D)
EIF4G1    D    240    Y    uc003fns.2    +    L    717    CTC    C    chr3    184039209    Exon 10
EIF4G1    D    240    Y    uc003fns.2    +    L    718    CTC    T    chr3    184039210    Exon 10
EIF4G1    D    240    Y    uc003fns.2    +    L    719    CTC    C    chr3    184039211    Exon 10

http://bioinformatics.mdanderson.org/main/Transvar

Introduction

TransVar is a reverse annotator for inferring genomic characterization(s) of mutations (e.g., chr3:178936091 G/A) from protein or cDNA annotation(s) (e.g., PIK3CA p.E545K or PIK3CA c.1633G>A). It is designed for resolving ambiguous mutation origins, arising from alternative splicing.

TransVar has the following features:

• supports HGVS nomenclature
• supports both left-alignment and right-alignment convention in reporting indels.
• supports annotation of a region based on a transcript dependent characterization
• supports single nucleotide variation (SNV), insertions and deletions (indels) and block substitutions
• supports mutations at both coding region and intronic/UTR regions
• supports transcript annotation from commonly-used databases such as Ensembl, NCBI RefSeq and GENCODE etc
• supports UniProt protein id as transcript id
• supports GRCh36, 37, 38
• functionality of forward annotation.

Citation: Zhou W, Chen T, Chong Z, Rohrdanz MA, Melott JM, Wakefield C, Zeng J, Weinstein JN, Meric-Bernstam F, Mills GB, Chen K. TransVar: a multi-level variant annotator for precision genomics. Nature Methods. In Press.

Have you tried the Ensembl REST API?

To convert from protein coordinates http://beta.rest.ensembl.org/documentation/info/assembly_translation

To convert from transcript coordinates

You'd need to get the Ensembl protein or transcript IDs, which you can get easily using BioMart (tutorial on BioMart here

Then, for your query, BRAF.p.V600E:c.1799T>A, input

From protein

http://beta.rest.ensembl.org/map/translation/ENSP00000288602/600..600?content-type=application/json

Output

{"mappings":[{"seq_region_name":"7","gap":0,"coord_system":"chromosome","strand":-1,"rank":0,"end":140453135,"start":140453137}]

From transcript

http://beta.rest.ensembl.org/map/cds/ENST00000288602/1799?content-type=application/json

{"mappings":[{"seq_region_name":"7","gap":0,"coord_system":"chromosome","strand":-1,"rank":0,"end":140453136,"start":140453136}]}

You can find the 3 b.p. codon which encodes the amino acid, but you need more information if you want single base pair resolution...

Can I use it if I don't know the amino acid position of the protein? I addition, as far as I understand, gene name and mutation info (either of the form c.123G>T or IVS4+1G>T) are not enough in order to deduce a specific genomic location as there may be more than a single transcript. Am I wrong?

Albeit built for mapping protein sequence intervals, my script "protein2genome.pl" can do this. It ships with the variant annotation tool CooVar.

Here is how it works. First you need a GFF or a GTF file with the coordinates of your genes. To test your specific example, I created a GTF file that contains only the first isoform of the BRAF gene, but you could also work with GFF/GTF files containing all human genes and isoforms or containing genes from any other organism. Then you can run protein2genome.pl like this:

echo "ENST00000288602 . . 600 600 . . . ID=V600E" | perl protein2genome.pl BRAF-001.gtf

Produces the output:

7    .    .    140453135    140453137    .    -    .    ID=V600E(ENST00000288602);segment=1of1;p_start=600;p_end=600

Explanation: I am basically piping a GFF-compliant input line specifying the ID of the transcript and the position of the protein sequence change into the script. The script then outputs the mapped genomic coordinates (chromosome 7, codon start=140453135, codon-end=140453137).

A more detailed explanation of input and output formats can be found here.

I would say this script is more useful for non-model organisms, because for model organisms with associated databases you have other possibilities to do that (see other answers in this thread, for example the Ensembl REST API which is quite neat).