If you generate a helluva a lot of data, you can get some fine-mapping structure of the genetic make-up of the inhabitants of the UK. This is what a group led by Peter Donnelly [before on The Blob] and Walter Bodmer from Oxford have done in their People of The British Isles PoBI study. Published in Nature 19 Mar 2015. They went through their data and isolated 2000+ people all of whose grandparents had been born within 80km of each other. That takes you back to about 1885 before rural depopulation sent everyone to hugger-mugger in urban stews and marry who-knows-who. They typed each of these people for half a million genetic variants aka SNPs and threw the whole mess into a huge computer program. For good measure they added equivalent data of 6000+ 'continentals' from Scandinavia, Germany, Belgium, France and Iberia and gave it all a damned good shake. The first outcome is that these data are extremely noisy, so that there is little-or-no predictive power - if you have a particular set of variants at these 500,000 genetic sites you won't have much confidence in placing your ancestors in a given tribe. You can't say you or anyone else has demonstrably Cornish genes or Yorkshire blood.
The third through tenth findings is in the detail of the periphery. The most distant outlier in England is Cornwall which is distinctively different from neighbouring Devon which forms a distinct [pale blue] cluster within 'England', as do the Welsh Marches and the West Riding of Yorkshire [that pale blue is not partic related to Devon's, they just ran out of colours]. Yorkshire is thus a political construct in the same way as Nigeria is, only with two genetically distinct tribes rather than 30. In a similar way the South and North of Wales are inhabited by quite different people - more different than Scots vs English for example.