donni preprint posted

May 25, 2023
Image
donni workflow

PhD student Linh Tran has developed a remarkably efficient approach to infer demographic history models from allele frequency spectra using feedforward neural networks. The preprint shows that this approach is just as accurate as likelihood optimization using dadi, while being instantaneous with a pre-trained model. To enable usage, we have pre-trained a wide selection of models, which our software donni will automatically download from the CyVerse DataCommons.