There has been some debate about the East Asian ancestry in Bengalis for decades. To me, the most parsimonious explanation 10 years ago is that it was mostly Munda. These are the Austro-Asiatic people of the highlands to the south and west of Bengal. There is also one Austro-Asiatic group to the north of Bangladesh, the Khasi.
I no longer believe this. I’ve looked at the genome-wide data and the signals into the Bangladeshis are much more like a donor population which is Tibeto-Burman. The Khasi in fact have more in common with their Tibeto-Burman neighbors than the Munda. At least genetically. This is one reason I am now leaning to the Munda maritime hypothesis, whereby the Munda actually landed on the coast of Odisha.
But there is a better smoking gun than genome-wide data. With a sample size of 700+ this 2011 paper did not identify any clearly Southeast Asian maternal haplogroups. This is probably an underestimate due to unresolved assignments, but it gives you a flavor. The majority of the Munda Y chromosomes are clearly Southeast Asian. The branch of O associated with Austro-Asiatic people. This 2018 paper using 240 Bangladeshis, with the largest samples coming from the Rangpur area in the northwest of the country, indicates a bit over 10% Southeast Asian haplogroups. This is in the range of the genome-wide admixture estimates.
It could be that in parts of West Bengal, to the south and west, the East Asian ancestry is Munda. But I am pretty skeptical, though willing to be proven wrong.
I do wish I had more non-Brahmin West Bengal samples though.
Note: I think the East Asian ancestry is probably a mix of various groups by the way. In the north clearly more Tibetan. In the southeast more Burman. The Khasi are clear vectors across much of Bangladesh.