An interesting conversation on Twitter on means for purifying DNA for PacBio and the risks of phenol-chloroform extractions restarted some pondering on the historical contingency of experimental techniques. Or, as the title says, why do we (today) purify genomic DNA the way we do?
I don't know much about the history of DNA purification, other than the small window of my own experience. I first purified DNA as a freshman lab project, and other than precipitating with cold ethanol I don't remember much. Might have used chloroform; don't remember any stark warnings about phenol. I do remember a bit of thrill seeing the DNA cloud the precipitation, and then the fun of spooling out gobs of DNA on a glass rod I had bent (any excuse for lampworking is one I'll take)
I did a lot of plasmid preps as an undergraduate using the boiling technique (I think; hazy on that). I do remember the unfortunate incident in which I left a tube of phenol out one evening, coming in the next day to find that it had melted through the tube and liquefied the styrofoam block I had the tube sitting in. Phenol was actually a familiar smell to me, as in those days of yore you could still buy sore throat spray that had a low concentration (and odor) of the stuff
When I was doing bacterial and fungal genomic preps in the early days of starbase, I used the modern spin columns. These are amazing, requiring almost no skill or technique -- though if you aren't careful the minifuge will shear the caps off your Eppendorf tubes, which annoys fastidious labmates. No risk of phenol burns, no nasty chemical waste and very small volumes, and the DNA is excellent for short read sequencing.
However, as we later needed high-molecular weight DNA for large insert libraries and PacBio sequencing, the old school phenol-chloroform was dusted off (not by me; I had gone all digital long since). These have all the above drawbacks, plus the concern that residual phenol carryover might damage downstream reactions. But as evident in the Twitter thread, phenol-chloroform tends to be the recommendation for PacBio reads -- though a quick skim of the MinION papers for long reads suggests that column methods are dominating that community.
How did phenol-chloroform become king? I truly don't know, and don't have anywhere near the time to find out. Sounds to me like an excellent undergraduate project, but I don't have any (let alone any to spare). But I can imagine that early workers in DNA work tried a number of different approaches, and found that phenol-chloroform (which used to be spiked with isoamyl alcohol, I don't know if it still is) worked well. Back in those days, disposal of solvents and lab safety weren't quite so worried about as today (my parents trained in chem/chemEng in the middle part of the previous century, and have told many a story). Presumably there were many failed attempts, and many avenues not very carefully explored.
In modern times, the cost and concern around toxic and dangerous chemicals has gone up, so now we choose methods that minimize or avoid these. But that still doesn't mean that every possible DNA extraction protocol has been explored; the space is simply too large. Now with the rise of long reads, a new set of pressures are appearing, plus the general trend towards larger numbers of smaller samples, meaning an emphasis on automation. Phenol-chloroform ain't automatable! Not unless you program a vision system to find the organic-aqueous interface.
If you want really high molecular weight DNA, then even phenol-chloroform isn't right. For techniques such as BioNanogenomics mapping or BAC library generation, the preferred method is to embed the cells in agarose plugs, and then digest the cells and clean them up in these plugs. I've seen it done -- very manual, but the steps are mostly washes requiring very little skill (I could do it!) and are run over days, not minutes or hours.
At the other extreme of easy DNA preparation, there are methods widely available online which use common and relatively safe household reagents to purify DNA. For example, one uses a blender, liquid dish detergent, meat tenderizer and alcohol. Another omits the blending and meat tenderizer but throws in salt. Now, I suspect these aren't the cleanest DNA preps and the blending would be expected to be a bit rough on the DNA (and the input amounts are huge), but they are described as generating DNA gooey enough to spool. The "Edwards" method in a paper I found appears to be a fancier version of the dishwashing liquid prep; it is compared against a prep using chloroform:isoamyl (aha!) but no phenol.
A final thought: perhaps the ultimate direction of this field is to eliminate bulk DNA preps altogether. With single-cell sequencing, a prep much like the agarose plug methods is used, but run in microfluidic devices or droplets in an automated manner. If these methods continue to develop so that they are reasonably inexpensive (and perhaps directly coupled to nanopore or other microfluidic-friendly sequencing technologies), perhaps nobody will still do bulk preps -- except to experience the thrill of spooling goopy DNA out of a solution!