My understanding of this is that the complete reference for molecular centralized evolution is The Selfish Gene itself. I'm going to cite Alan Grafen's essay here,
Richard Dawkins, proposes that the logical, fundamental unit of evolution is the replicator.
The replicator is the fundamental unit of reproduction - whether these replicators survive or do not defines the properties of the biological system they reside in. These words were chosen carefully because DNA sequences themselves are not replicators exactly - many individual copies make up the animals in a gene pool and the replication of one copy or another doesn't matter. The distinction between a physical strand of DNA and the replicator is the bulk of more than one of Dawkins book. It shows why the life of an individual means nothing to biology or evolution, how the boundaries of genes as transcriptional elements do not define the replicon (commonly misunderstood by molecular biologists). The replicon (a genetic unit of replication) manifests a behavior or phenotype in the organism that helps it reproduce and is reproduced. This can be cooperation with other members of your species or egg rolling or killing other sperm and is usually the result of many individual mRNA transcripts working in concert. A given gene as defined by mRNA transcription commonly participates in multiple replicon type behaviors.
But DNA is in the mind of the molecular biology camp highly intertwined as a replicator because it is the physical medium through which the replicator is passed from one generation to another. Without DNA, meiosis and the rest there are no replicons. The earliest replicators were possibly simple transcripts which self replicated and not much more, but over time they have become a network of intertwined molecules working in a system.
To my understanding this is axiom more than proof. It does not need to be proven because the physical transmission of DNA replication over time is the minimal necessary scenario to picture biology and evolution as we understand it. The selection amongst groups of genes or groups of cells or groups of individual multicellular animals seem unnecessary given the physical framework of DNA replication.
Certainly that these connections have been shown. The inspiration for The Selfish Gene is the confirmation of Robert Triver's Selfish Gene hypothesis with the discovery of t a terribly selfish gene which takes up the better part of chromosome in drosophila and increases its chances of showing up in the next generation by showing up in sperm more than half the time. There are lots of other examples in the Selfish Gene and the Extended Phenotype, which answers many of the common questions about this thesis. The literature since has been pretty productive. That being understood, its one thing to prove that something is true in biology, and another thing to prove that all other similar mechanisms are simply not true.
Yet even in Dawkins essays we see that the replicator is a bit of an abstraction some practically indefinable group of genes. Even as they each compete against each other, they create phenotypes collectively and the replicon becomes abstracted from the fundamental mechanism of replication. The most popular example of replication Dawkins creates - the meme - has no relationship to DNA at all - a sticky idea that spreads socially and replicates in the minds of individuals.
When we look at terribly complicated but obviously advantageous behaviors, the connection between the replicator and the competition of individual genes gets muddy. EO Wilson, having discovered pheromones and started ecology and sociobiology as disciplines still publishes group selection arguments. Having studied social behavior in insects over his career and co-authored one of the most complete compendiums on the subject, he has put forth consistant arguments since the 70s that group selection still has a place in evolution. Unlike others, Wilson has I think always kept the conversation scientific and isn't jumping off any ideological cliffs (though responses say differently, the whole conversation is still published).
I don't see how an unbiased forum can ignore the fact that the editors of Nature still call this an open discussion. Some may object, but calling the subject closed is overreaching I feel. There are others besides Wilson; I'm not familiar enough with the whole discussion to go into David Sloane Wilson and the rest, though they publish in reputable places.
I've started in this post since the question has been sitting around for a while. I expect others will post their corrections/objections/views - please do! Whenever group selection is mentioned it often happens that someone will pop up and say 'this is not appropriate for a scientific discussion of biology here' but never any justification of disqualifying questions - I think that's not really scientific. The fundamental attraction to academic thought to me is that minority voices that have legitimate observations are tolerated and often result in scientific revolutions, after which they are venerated for their persistence.
Here's a good place to tell us why some questions are beneath discussion in biology. I personally do not agree; even if a position is intertwined by politics and religion, its the scientific value that should be judged, not by the company it keeps - lest science become just another intolerant clot, not driven by logic, observation and facts. I ask you, why bother then?