peterzzq / velosim Goto Github PK

Hello all!

First of all, congrats on the publication of your VeloSim preprint. As the author of dyngen, I thought it was an interesting read; you definitely bring up interesting points of dicussion. For example, the fact that VeloSim includes a two-state kinetic model, it's something we've been considering including in dyngen as well, but in the end thought it would change the expected behaviour of dyngen too much.

I don't want to go into the benefits or downsides of VeloSim vs. dyngen at this stage, but would like to briefly discuss a statement made about dyngen:

Dyngen and SERGIO are able to simulate both spliced and unspliced mRNA counts of cells in a continuous dynamic process, but they each has its own limitations in practice. Dyngen provides predefined trajectory structure types in the package, such as bifurcating or cycle, but it is not flexible enough for more customized trajectory structure where user need to specify the exact developmental tree structure and the length of each branch.

While it is true that dyngen provides a lot of predefined trajectory structures (documented here), users are also allowed to specify a custom trajectory topology (documented here).

In your next version of the preprint, could you have this statement changed?

Kind regards,
Robrecht

There is one small mistake in your source code of simulate_counts.R.

• In function SimulateVeloCycle, the 455th line should be return(list(counts_u = result$counts_u, counts_s=result$counts_s, kinet_params=final_kinetics, state_mat=result$state_mat, cell_time=result$cell_time, velocity=result$velocity, dynamics=dynamics, backbone = evf_res[[2]]$pop, chosen_hge = chosen_hge)), you mistook counts_s=result$counts_s as counts_s=result$counts_u.
• The same mistake is also in the 577th line.
  Thanks for you package, this helps me a lot! Please check it.