Where do NGS projects most often Fail?

[cystron]

When thinking about a new project, where do you see the most failures?
1. DNA/RNA extraction
2. Library Prep
3. Sequencing Run
4. Alignment / Assembly
5. Secondary/Tertiary analysis and BioInformatics

At what point do you finally feel like a project is "safe"?

[GenoMax]

I am not sure if this question has a generally applicable answer. Projects can fail at any step mentioned in your list. A project can never be "safe" (to borrow your phrase) until the data is analyzed and the paper written :-)

[gringer]

Experimental design: not choosing appropriate controls, too few biological replicates, or doing sequencing on the wrong cell population (e.g. whole-blood RNASeq for an obesity study).

[Genetic Librarian]

Experimental design: not choosing appropriate controls, too few biological replicates, or doing sequencing on the wrong cell population (e.g. whole-blood RNASeq for an obesity study).

This!

Biggest mistake is too few replicates. Individual samples can drop out for a variety of reasons, so easily your glorious n=3 study has treatments with only 2 replicates in them. And you can not easily repeat the missing samples without having a potential batch effect confounder.

Also, from my experience as a core lab head, usually it is "**** in, **** out", meaning most library preps / sequencing runs fail because input material was not ideal (but customer wanted to try anyway after being contacted following QC).

[kwaraska]

To second what another poster said.....there is no safe place until you have your data analyzed but the most common problem is garbage in,garbage out. In the core lab I was in, we had similar issues where a customer would insist we use their sub-optimal ample, and the data would commonly be garbage.

It is worth checking all samples on a Bioanalyzer or Tape Station type instrument to really look at what you are dealing with. The most common issue would be someone using a nanodrop to get a concentration, but upon TS we would find it was less than 50% of what they reported