Improving Accessibility and Confidence Assessment of the TT-Method

Sjödin et al. 2021 introduced the TT Method for calculating divergence times between two populations using two genome copies (i.e. one diploid individual) from each of the populations, and the ancestral states for each position. The method can reduce bias by conditioning on an outgroup genome, called the TTo Method. However, after releasing their Python pipeline, the authors received several user issues and queries. They were also interested in evaluating the chosen method for confidence assessment: Weighted Block Jackknife (WBJ). Therefore, this project aimed to investigate the confidence assessment and make the published scripts more user-friendly. The effect of changing the window sizes for WBJ was evaluated using genome sequences from the 1000 genomes project: one Finnish, one Japanese and a Yoruba individual as the outgroup. The speed of the pipeline was recorded with each run to see if window sizes had any effect and compared to the previous method. The WBJ means were found to fluctuate at window sizes >10Mbp, being affected by bias within the genome. While smaller window sizes, including the previous default of 5Mbp, did not show this bias, window size did not effect the runtime, therefore a smaller default window size of 2.5Mbp is now suggested as it brings the WBJ mean closer to the whole genome estimates. The scripts were re-structured, parallelized on CPUs and providing input was moved to the command line. The user manual was also updated to include more advice on how to run the pipeline and produce the required ancestral states file. The runtime of TT, TTo with TT counts provided and both TT and TTo at the same time was recorded (31, 44 and 57 minutes, respectively). This was 14, 17 and 48 minutes faster than their previous counterparts. These improvements to the pipeline should make it more viable for future users as a method for obtaining divergence times between populations with genomic data.