the human reference genome is "haploid" meaning that, for example GRCh38.p14 has only one copy of chr1-22,X,Y. you would have two copies of the gene in each diploid cell in your body, but for bioinformatics, we only "map to a haploid reference". this answer is a little insufficient, so i'll try to explain where mom and dad might come in:
let's say we "sequenced your genome". if you sequenced your genome, you would get "reads" from the sequencer machine, which are like millions of 150bp sequences from illumina, in a FASTQ file.
what do you do to analyze your sequence data? well, you would map your reads versus a reference genome, the reference genome being like GRCh38.p14 there.
this mapping program, like bwa, or bowtie, or many others, produces a BAM or CRAM or SAM file, and it can tell during mapping that it should map at some particular position, but maybe there is a 1 letter difference. this could be a "SNP"
then, maybe, only half the reads at a particular position have a SNP, and the other half don't. since approximately half the reads come from your mom and the other half came from your dad, that means maybe you inherited a SNP from your mom (which is shown on half the reads) but didn't inherit a SNP from your dad
this shows how we can get information about your diploid genome, even when we are comparing to a haploid reference. you do ask, well, shouldn't it be at two different start positions? well, we assume it only has one start position when we compare to the reference genome, and we only look at mutations to see if those mutations affect the gene of interest.
in the future, "graph based genomes" may be used instead of a single reference genome. graph based genomes is kind of like mapping to multiple reference genomes at once, but it is a somewhat advanced topic that is still evolving, so comparing to a single reference genome is commonly done today.
hope that helps