I think this explanation is missing the real explanation of why CRISPR is useful.<p>I want to take a crack at it. Let's say we want to change in the genome the sequence (where each of the 'letters' represents a somewhat long stretch of base pairs):<p>ABCDE to ABC'DE<p>you would normally create the sequence<p>BC'D<p>in vitro and put it into the cells. The organisms contain mechanisms to match the B & D sections and thus 'swap out' the C section for the C' section.<p>Note that C could be "" which would make the process a straight insertion. C' could be "" which would make the process a straight deletion. C and C' could be a single base pair, which would mimic a point mutation, etc...<p>However, you don't have TOTAL control over this process, it's stochastic, and doesn't have 100% efficiency. So you have to do something clever to make sure you have what you want. Typically that involves inserting resistance to a chemical factor (e.g. antibiotic). So for insertions (if you don't mind a dirty insertion) it's fine, but for other transformations like mutations and deletions, you might have to be clever, and say, do C -> C' -> C'' where the C' includes the selection factor. And C'' is chosen either because it lacks a toxic factor that we put in alongside C' or by doing a reverse selection where we pick clones and test to see if they die (and keep some of the originals in case they pass the test).<p>This process generally works quite well in most microbes with small genomes (E. coli requires a tweak to the process). It is basically effortless with yeast.<p>With higher eukaryotes it's not quite so simple. A competing process is inserting the BC'D sequence <i>elsewhere in the genome</i>. It's not entirely clear why this is such a huge problem, but likely it's because of the increasing complexity and size of the genome. If C' contains a selectable marker, it becomes difficult to distinguish between what you want (ABC'DE) and just BC'D somewhere random in your genome. Both are resistant. And the process becomes bogged down by the need to isolate single cells, propagate them, and check to see if your strain has the substitution you want (relative easy, just a PCR reaction) and no other substitutions elsewhere in the genome (haaaaaaard).<p>The CRISPR advantage is that just before you add BC'D to your cell you create a scission somewhere in C so you're left with ABc//cDE - and what this does is triggers the cell repair system to search for B & D sequences to hook into. Naturally it will find BC'D. Well, if it doesnt, usually a fragmented chromosome will also result in death of the cell, so you're virtually guaranteed that the surviving cells have ABC'DE. With this, the rate of successful targetting so exceeds the rate of random insertion that the necessity to check is basically eliminated (or at least you don't have to search through so many clones to pull out a total success).<p>The net effect is that for many higher organisms genetic manipulation becomes much much much easier. YMM(still)V with some plants which have high level of repeats within the genome.