Why are the results different between memory=0 and memory=1?

[JinyongChung]

Hi, I am really interested in your great work. I am analyzing my data using your algorithm and I have a question.

From my short understanding, the results of memory=0 and memory=1 (in your run_VIA() code) should be the same. But the results are not.

Could you explain why?

Thanks.

[JinyongChung]

Oh, I repeated the same analysis and found that they can be the same with a little variance due to the inherent randomness of the algorithm.

So theoretically they are the same, right?

Thanks.

[ShobiStassen]

Hi Jinyong,

The results for memory = 0 and memory = 1 are theoretically the same. However in the memory = 0 case, we default back to the Via 1.0 mode to avoid the extra computational load of using the second order walks function where the edge weights would be recomputed and tracked even though they are just be being multiplied by a factor of 1.
small differences are due to the inherent randomness of the method.
I need to tidy up the comments around this in the code and function descriptions, and will aim to do this in a couple of weeks when I am back in the office.
Thanks for your question :) feel free to check out the paper on Genome Biology