If you want to detect SNPs on a specific region I advise you to do a SNP calling on all genomes (aligning reads with a mapper like bwa, then use a variant caller, as freebayes).

Then you can extract the SNPs of the specific region you are interested in with vcftools.

If your FASTA has metadata in its record header that points to its location on the genome, you can use that directly to map any SNPs to it via BEDOPS bedmap and vcf2bed:

$ echo -e 'chr2\t2241384\t2244383' | bedmap --echo --echo-map-id-uniq --delim '\t' - <(vcf2bed < snps.vcf) > answer.bed

Replace chr2 with 2, depending on the format of chromosome name in your snps.vcf file. This could either be UCSC (chr2) or Ensembl (2), most likely.

The file answer.bed will contain the 3k nt interval and a listing of all SNP ID values that map to — or associate with, or overlap — that interval.

If you don't have metadata in the FASTA header that tells you where you are, you could use a BLAST search on the sequence to get back the location of your sequence for your genome of interest.

Then you run the command above, again replacing what goes into echo with whatever region comes out of the BLAST search.

I would simply use GATK HaplotypeCaller with -L option