Does changing the # of cores or memory change the number of genes and transcripts assembled by Trinity?

Question

[john]

We use the same fastq file and run it multiple times on different specs to see what the optimal number of cpu and memory is for speed but the thing is we notice that the number of transcripts and genes assembled varied each time (+/- 20). Is it because of the number of CPU and memory we use or is it just because of Trinity's algorithm? Thanks.

Answer 1

[ricohagelaar]

If I recall it correctly trinity uses a different starting point for the assembly. This may explain the difference in the number of assembled transcripts/genes. The CPU and memory should not give a difference.

[john]

Thanks, I was trying to figure out yesterday where exactly this happens in the algorithm. In choosing the seed k-mer for inchworm perhaps?

[ricohagelaar]

I am sorry, I didn't use trinity that often to know where it might happen. However I know for sure that it shouldn't be caused by the number of cores/memory.