6.2. Why is the progress bar for my assembly stuck at 78%?
For users of the Genomics Workbench, there are three phases when running a de novo assembly: a pre-processing phase, a computational phase and a post-processing phase. The pre- and post-processing phases are part of a java process. The computational phase is run via a native binary.
During the computational phase, there are several stages. The final stage is that of optimizing the de Bruijn graph. Up to that stage, the computational phase is able to run on multiple cores. This final, optimization stage however, currently runs only on 1 core.
For an analysis where Simple Contig output has been chosen, you will likely see this final stage start when the Workbench progress bar is at approximately 78%. The length of this optimization phase depends greatly on the quality of the data and the complexity of the genome being assembled. This phase can take a long time, especially in comparison to the amount of time that was taken up to that point.
The progress bar is, unfortunately, not very sensitive to the progression of this particular phase of the analysis. The lack of change to the progress bar (e.g. seemingly hanging at, say 78% or 80%) is normal and does not mean the analysis has stopped. (The percentage progress reported in the Workbench should be interpreted more a statement about what stage of the analysis you are on than an estimate of how much time it will take before the analysis is complete.)
Our advice would be to leave the analysis running. Assuming no other problems later on, the analysis should finish normally.
Our developers are aware of the fact that the graph optimization phase would benefit from being able to be run on multiple cores and this is something they plan to work on in future. Associated with this, we hope the progress bar will also be worked on such that it provides better feedback on the progression of the analysis.