Although it is unusual for an undergraduate to get publications, it does
happen. My former advisor at Michigan was (at the time) the Neuroscience
chair, and shared with me statistics from MD/PhDs at Michigan. All of them
had the qualifications I described. My guess is that this is the case
at all top medical schools (i.e., Michigan, Stanford, UCs, Harvard, JH,
etc.). However, it maybe easier to get into an MD-PhD program at
a 'lesser' (which is usually only perceived not actual) medical school.
However, I would choose a good PhD program over a lesser MD-PhD program,
since the qualifications for PhD programs are easier.
In any case, publications are certainly not necessary for admission into
a PhD program, but independent research certainly is, at least for the
top research universities. Of the 22 students in my class at UCLA, I know
of only one who did not have significant independent research.
Canned laboratory classes are almost always a waste of time. You should
contact a faculty member at your university whose work interests you and
arrange to volunteer in their laboratory. If one won't take you, the next
one will. At most universities you can get credit for the work as
some sort of 'independent research,' and this will lead to the letters
which will get you into grad school.
So basically, this is the way I'd put it: for top research schools,
you can get in with just good research and great letters, even if your
grades aren't stellar. For some schools (e.g., Penn, or MIT), you
need both great letters, research, and great grades and GREs. This is
also usually the case with MD-PhD programs.
In any case, the most important thing is that you find a faculty member
whose research interests you AND whose students say he/she is a good
advisor. If you find this, the school is actually a matter of taste and