Read on Twitter
Lots of pre- & postdoc grant deadlines coming up (I think @NIH should delay, but I digress...) so I know there are many grad students and postdocs working (very hard & often in not great circumstances) on these applications. So here are a few tips & tricks that go beyond...
...the standard stuff like 'have a Sr. co-mentor w/ funding', 'have a strong training component', 'give it to peers and faculty to read', 'have multiple 1st author papers'...some you can control (get feedback), some you can't (multiple papers) so here is advice EVERYONE can use.
Frame your question in terms of a model (a graphical hypothesis), have that model as a figure, can overlay how aim is testing aspects of that model. Reviewers need help, this can be a summary they can refer back to as they review your grant. Here are a few of mine...
Give 15-20 lines to 'Potential outcomes, interpretations and future directions'. If you get Result X, what does that mean in context of what is known? What would be a potential follow-up experiment? How does this connect with results from your other aims? What if you find...
...the opposite? Why is that important? What are the alternative ‘players’ in your model you could test in the same system? In training grants it can be difficult to balance all the technical/experimental details expected by reviewers with the ‘potential outcomes’. But...
...this is where to show you are thinking like a scientist, synthesizing different results to set up the next step in testing the model. Discovery does not end with a single exp. outcome, it is just the beginning! On avg. I dedicate 30 lines of text PER AIM to this in my R grants
Pitfalls and alternative approaches. Point out some potential technical problems (better you write it than the reviewer think it…) and what alternative strategies you have at your disposal. Different cell isolation methods, use a cell line instead of primaries...
...use a different cre line, different reporter, etc.…let the reviewer know you have thought through what might not go well and you have a reasonable plan in place to STILL test your hypothesis. This is important for practical purposes, too 'cause science is crazy fickle...
References. Do not rely too heavily on reviews, cite the primary literature, especially if essential for your model, this demonstrates scholarship. Also, don’t cite things unless you have given them a reasonable read through. As a reviewer, I will look up cited references...
...and I find that applicants sometimes cite papers that do not support the statement in the text. Can suggest they are 'reference padding' instead of using the literature to build support for the hypothesis. This is worrisome to see in a training application.
Use your figure legends to your advantage - if you have a graph with replicates, list out the statistical analysis that was performed, replicates, p-values, etc to demonstrate rigor in pilot data collection. Figure legend text should be similar size to the main text...
...or have a background to provide contrast if print is a little smaller. For figures with qualitative microscopy images, make sure images are not fuzzy or dark (don't forget scale bars!!), especially important to check after Word file is made into a PDF. Try several different...
...file types (TIF is best in my experience, DO NOT USE JPEG!!), DPI (200-250 DPI), zoom in/out on the converted PDF, if possibly check on different screens. I spend a lot of time on this because if you are going to have microscopy data in there vs text, make it worth 1000 words!
Ok, last is format tricks. If over pg limit & can't trim back more text or need space for model figure (always trim back extra text before this step!), in Word, select all your text, go to 'Line paragraph spacing' < 'Line spacing options' < set 'Line spacing' to 'Exactly' 12pt
Always count lines to make sure you are still within the limits for the application (this is totally fine for @NIH) but beware this 'squishes' the text, which can bother some reviewers (I am not bothered by it, despite heavy 
use my eyesight is 20:20) so this is a last resort
use my eyesight is 20:20) so this is a last resort
One tip to help with 'squish' is to add 3-6pt space in between all paragraphs, doesn't impact line spacing as much and is a little easier on the reader.
I have learned all this from years of trial and error, mistakes, grants ND...basically a combo of failure and feedback/help from more senior scientists. And I am still learning, every grant I try something different. Good luck w/ grant writing, I hope some of this is helpful!
