That type of plot is called Sequence Logo. The X axis represent the nucleotide positions of a sequence, and the Y the entropy of each possible nucleotide. The size of each letter indicates how frequently it is found at that position: for example position 8 is almost always an A, while position 5 can be a T or C.

You can create sequence logos from a web interface using a popular application called WebLogo, or using R packages such as SequenceLogo.

The plot shows the comparison of the reference human sequence with N-number of samples (patients). The samples turn to have variation of sequences in that region. Now from these samples, you can compare the probability how often a given nucleotide is mismatched/mutated in sapmles. If the letter is big it means the sequence is invariant.

You can use weblogo either in their webpage or in R to make that figure.

Or if you have Chip-seq peaks, you can do motif analysis and make those plots using HOMER tools.

Here is motiflogo, which can make a SNP-specific motif logo representation. See figure, please.