Be cautious in discounting the state of Texas angle. 1800s census records are notorious for having wrong states of birth and repeating the error over decades. H.N. Sheeler was 14 in 1870, and G.J. Sheeler was 10 that same year, and a letter says the family was residing in Texas up to the time of the parent's death, and a birth record says G.J. was born in the state of Texas. This could mean the family was residing in Missouri up until around 1859 with H.N. being born there ca. 1856, moved to Texas where G.J. was born ca. 1860, and then the parents died a year or two or three later, after which relatives brought the very young boys back to Missouri where they grew up. G.J. would have, except for just a brief time in his infancy, been raised in Missouri. As census enumerators went door to door in the 1800s and G.J. was off working, the wife or child or whoever in the household answered the enumerator's questions might very well have reported that he was from Missouri. The place of birth for my own great great grandfather changes from Virginia to North Carolina from census to census. I think it all depends on who was answering the census enumerator's question.

Anyway, Missouri is certainly a possibility, but I believe Texas no less of a possibility given the information you have provided. My advice is do not eliminate either possibility, do not let theories eliminate avenues of research, and do not be afraid to discard theories and develop new ones....

What does the state of Texas census for 1860 indicate for Sheelers? Perhaps G.J. was not yet born at the time it was taken, so maybe he wouldn't be on it but 4-year-old H.N. would. Could be the family was en route to Texas and won't show up on the 1860 Missouri census or Texas census either one.

Texas was notorious for politically-inspired mass executions in the early days of the war. Perhaps the Sheelers were caught up in one of them (however, women were not reported to have been targets so I am very leery of this possibility). The mass executions are pretty well documented, so you might look into them. They politically-inspired murder of a woman and mother of two would also likely be documented, whether it occurred in Missouri or Texas.

And, given the fact that women just weren't victims of internecine murder during the war, maybe, very possibly, the deaths had absolutely nothing to do with the war. Perhaps the parents simply died of disease during the time period, which started out as "they died during the war," and evolved into "they died in the war...." In that vein, there could be a whole host of causes of death that were unrelated to war. In any event, good luck in your search.