Totally depends on the track w/in the institution. Though do note that for some AMCs the only way to get on faculty is to have a grant and it's highly advantageous to do post-doc at the institution you want to be faculty at. For the research-based tracks at the AMCs I've been at --- A time of appointment assistant profs typically have 15+ publications (> 1/3 FA) + a NIH equivalent grant (PI K series grant or Co-I on others grants). Promotion to associate for these research-based tracks (both tenure and non-tenure), are ~ at least 1 R01s as PI (or equivalent) + a few smaller research grants (R21, R34 etc) as PI and/or several Co-I R01s. 2 R01's if you want to be completely safe for promotion. It's grants that matter and publications are largely secondary (though you of course have to have published, > 30, 1/2 FA though many have many many more). Clinical tracks have different requirements, some have high research expectations others may have no research expectations provided you are able to establish national/international recognition in other ways.
Based on my friends publications #s are higher at university-based R1s and grant expectations may be lower (albeit definitely encouraged). My friends at R1s were initially appointed with publication numbers in the low 40's (though this can be much less if their research high impact and/or is more basic/neuroscience-focused or limited to clinical trials-- basically things that take a lot of time and resources to do) and as associate profs are hitting 100+. It's easier to publish a lot if you have motivated students doing a lot of the work. 4-5 students each aiming for 1-3 papers a year ..... and it's a great way to teach and get work out there.
In terms of what looks good a pre-doc NRSA (F31) is definitely a plus, though the majority of faculty I have worked with at AMCs actually did not have one in grad school. IMHO it's more important to develop expertise in a particular area and innovate because then people will want to collaborate w/ you.