Hi, Donna. I'm afraid this isn't going to be thorough answer (long, but not thorough), and there will still be a hefty dose of jargon in it...kinda unavoidable.
From everything described so far, I gather William H. McDonald (Mcdonald-1051) is the orphaned profile you suspect might be your brick wall. If I'm not correct about that, then...never mind.
But for that William H. McDonald there are only yDNA test-takers shown as possible connections, six of them. You can't compare your DNA to those results shown, and would have to contact the men and ask whether they've taken autosomal DNA tests and, if so, where you can locate their results.
However... From the yDNA information alone we can tell that there are least two errors in the WikiTree pedigree involving William H. McDonald. Of the six test-takers shown in the "DNA Connections" panel (and speaking solely of the patrilineal line yDNA): one of them cannot be related to any of the others; two of them (managed by the same person) can be related only to each other; the three remaining likely are related to each other but more information would be needed to make that a solid determination.
Confusing enough? The upshot is, because the yDNA is not in agreement, it's not going to be possible to use DNA in this case until the genealogy is better researched and the paper trails corrected.
The jargon-filled specifics next, but a quick preamble. What we can tell about these yDNA test-takers is only from the haplogroup information they posted on their profiles. Like it sounds, a haplogroup (a term applied only to uniparental DNA: yDNA and mtDNA) deals with genetic markers that can effectively identify populations in historical groupings, often in terms of anthropological rather than genealogical timeframes. The markers used, single nucleotide polymorphisms (SNPs), behave differently in yDNA and mtDNA than they do in our autosomal DNA. In uniparental DNA they never recombine, which means that changes happen only by mutation.
And we've learned that many of these mutations can be treated as hierarchical: we know some of them are much older than others, and over the course of the last three decades of DNA testing it's been possible to develop a haplotree, a branching structure similar to our family trees with the oldest known mutations at the top, and more recent ones filling in widening branches below. The (mostly) full sequencing of the Y Chromosome by the Big Y-700 test at Family Tree DNA has rapidly expanded the number of identified branches on that haplotree. In September 2018 there were 16,361 identified branches; today there are 45,901.
At the very deepest levels there may be only a handful of men who have been found to be of the same haplogroup as designated by what's called a "terminal SNP" (an older term that's a bit misleading: before the haplotree started branching so rapidly, "terminal" sort of applied, but now a man might be classified one additional level deeper at any time as new test results are evaluated). In those cases, the haplogroup might provide some measure of information about the degree of relationship between two test takers; it might provide positive evidence.
Higher in the haplotree--as the mutations in the hierarchy become older--the haplogroup can't offer any strong positive evidence for genealogy, but it's extremely useful as negating evidence to show that two men do not share a common patrilineal ancestor in the genealogical timeframe. And that's what we have on William H. McDonald's profile.
I'm going to refer to the haplogroup designations as shown in the "DNA Connections" pane on the profile so that I don't name the WikiTreers; you may need to do a little hunting to follow along.
The entry showing haplogroup R-CLD27 has probably done a Big Y test. That's a deep branch, or subclade, under the R-M198 haplogroup; the older, long-form name is R1a1a. The same person also manages the entry under that one, a 67-marker test showing haplogroup R-M198. I would guess those two men to be closely related.
Entry number six at the bottom of the list is, likewise, shown by testing to be estimated as R-M198. He has tested only to 37 STR (short tandem repeat) markers, but both he and the first on the list (R-CLD27) have added their data to MitoYDNA so we can at least get a comparative glimpse. Via the infinite allele model that FTDNA uses, at 37 markers they would be considered a genetic distance of 2; so they're very possibly related in the genealogical timeframe. Not enough data to make an unqualified decision, but my guess is that we would find that entries one, two, and six share a male ancestor within a time period that would include William H. McDonald.
Entries three and four seem likewise to be related to each other. But there's a bit of a problem here. R-L193 is a deprecated haplotree designation; it doesn't signify a branch on the trees either at FTDNA or YFull. As best I can tell, the current branch designation that includes L193 as a synonym or associated SNP is R-S5982. This has the higher, and more common, R-M269 (or R1b1a1b) as a parent branch. R-M269 and R-M198 haven't shared a common ancestor for roughly 22,000 years. So this pair cannot be related to the three R1a1a folks.
Number five is a bit of an outlier. R-A940 also descends from R-M269, but from R-DF13 A940 branches to R-DF21 while L193 branches to R-L513. DF21 formed about 4,100 years ago, so that means R-A940 is not genealogically related to any of the other five test-takers shown on William McDonald's profile.
A long answer, but I'm afraid one that can't offer any real help for sorting out McDonald-1051. Bottom line is the DNA information does us no good until we can figure out where the errors are in the lineages associated with William H. McDonald.