The bottom line
Y-DNA origins in North India are favored with a 65.6% probability of that subregion being the place of origin, and mtDNA origins in Northwest India are favored with a 71.3% probability of that region being the place of origin. Both of these probabilities are much greater than the runner up and are consistent with each other.
East India is the runner up for mtDNA (19.7%); but, the probability of an East India origin for Roma Y-DNA is only 2.7%.
The runner up options for Y-DNA origins are Central India (19.0%), and South India (10.3%). The probability of an mtDNA origin in South India is just 6.1% (all of which is from Southeast India as opposed to Southwest India). There is no separate Central India estimate for the mtDNA data which are broken up into somewhat different regional bins.
Background and analysis
For the most part, this confirms prior genetic and linguistic research on the issue, which is quite substantial, although some prior research has suggested a migration of men from elsewhere in South Asia to Northwest India, followed by marriage of local women, and a second migration from there to Europe.
I've previously blogged research on this topic on May 12, 2012, January 11, 2011, and August 31, 2009. There is also a passing reference to mtDNA U6 in Roma people here.
This study doesn't rule out these more complex scenarios, but does suggest that a more parsimonious theory is sufficient to explain the uniparental genetic data.
The paper and its abstract
The Roma, also known as ‘Gypsies’, represent the largest and the most widespread ethnic minority of Europe. There is increasing evidence, based on linguistic, anthropological and genetic data, to suggest that they originated from the Indian subcontinent, with subsequent bottlenecks and undetermined gene flow from/to hosting populations during their diaspora. Further support comes from the presence of Indian uniparentally inherited lineages, such as mitochondrial DNA M and Y-chromosome H haplogroups, in a significant number of Roma individuals. However, the limited resolution of most genetic studies so far, together with the restriction of the samples used, have prevented the detection of other non-Indian founder lineages that might have been present in the proto-Roma population.Martinez-Cruz, et al., "Origins, admixture and founder lineages in European Roma" European Journal of Human Genetics (16 September 2015) doi:10.1038/ejhg.2015.201
We performed a high-resolution study of the uniparental genomes of 753 Roma and 984 non-Roma hosting European individuals. Roma groups show lower genetic diversity and high heterogeneity compared with non-Roma samples as a result of lower effective population size and extensive drift, consistent with a series of bottlenecks during their diaspora. We found a set of founder lineages, present in the Roma and virtually absent in the non-Roma, for the maternal (H7, J1b3, J1c1, M18, M35b, M5a1, U3, and X2d) and paternal (I-P259, J-M92, and J-M67) genomes. This lineage classification allows us to identify extensive gene flow from non-Roma to Roma groups, whereas the opposite pattern, although not negligible, is substantially lower (up to 6.3%). Finally, the exact haplotype matching analysis of both uniparental lineages consistently points to a Northwestern origin of the proto-Roma population within the Indian subcontinent.
Y-DNA I-P259 is also known as I1c and as I-M507 and is currently considered a "private" haplogroup exclusive to one individual, surname or group of closely related individuals.
Y-DNA J-M67 is also known as J2a1b and J-M92 is also known as J2a1b1.
A paper at the ASHG 2015 Conference also addresses this subject and is less definitive in its conclusions: B. Melegh, et al., "Refining the South Asian origin of the Roma people."
Purpose: Historical and linguistic studies have suggested that Roma people, living mainly in Europe, migrated into the continent from South Asia about 1000-1500 years ago. Genetic studies, based on the examination of Y chromosome and mitochondrial DNA data, confirmed these findings. Recent genetic studies based on genome-wide Single Nucleotide Polymorphism (SNP) data further investigated the history of Roma and, among many other findings, suggested that the source of South Asian ancestry in Roma originates mainly form the Northwest region of India.
Methods: In this study, using also genome-wide SNP data, we attempted to refine these findings using significantly larger amount of European Roma samples. We also had the opportunity to use more data of distinct Indian ethnic groups, which provided us a higher resolution of the Indian population. The study uses several ancestry estimation methods based on the algorithmic method principal component analysis and model-based methods that apply Bayesian approach and uses Markov chain Monte Carlo or maximum likelihood estimation.
Results: According to our analyses, Roma showed significant common ancestry with Indian ethnic groups of Jammu and Kashmir, Punjab, Rajasthan, Gujarat, Uttarakhand states, e.g. with Kashmiri Pandit, Punjabi, Meghawal, Gujarati and Tharu. However, we found strong common ancestry with Pashtun and Sindhi, ethnic groups living in Pakistan. Populations of Northeast India have also strong common ancestry with Roma. These ethnic groups are Brahmin, Kshatriya, Vaish.
Conclusion: We can conclude, that Northwest India plays an important role in the South Asian ancestry of Roma, but they have similarly strong ancestry with some Pakistani ethnic groups and we can find populations in the east region of North India, which also could function as a source of Indian ancestry of Roma. However, ethnic groups of the southern region of India do not show strong relationship with Roma people, living in Europe.