assembly times using lrasm

[Scott]

Hello,

I am using the lrasm pipeline to assemble a 5.4 Mb bacterial genome. According to contigs.log, minimap2 took 19 min to complete.  Racon was then started and has been running for 28 hrs using 7 threads.  Can anyone provide a general guideline for how long assembly should take (the computer running EDGE has 2 Intel Xeon E5-2620 v2 Ivy Bridge_EP 2.1 GHz processors (total of 12 cores) and 256GB ram)?  For comparison, the same data set was assembled using Unicycler which also used miniasm/minimap2 and racon.  Minasm took 15 min to assemble and racon polishing (9 rounds) took ~4.5 hrs using 8 threads.  The depth of coverage for the genome is ~250.

Thanks,

Scott 

[Lo, Chien-Chi]

Hi Scott,

 

Is the run ever finished? We don’t have many nanopore use cases. Here are two cases we recently ran. The Ecoli (4.6 Mb) dataset with ~60x depth coverage took 1 hrs 06 mins using 8 CPUs.  Another 2Mb genome with 230x depth coverage took ~10 hours using 8 CPUs. The more depth coverage does affect the assembly time. In addition, by default, EDGE uses minimap2/racon to do reads correction before miniasm assembly in the lrasm pipeline.  We found the reads error correction step is usually the time bottleneck of the pipeline. You may try to turn the reads error correction off and see how it goes.

 

Thanks,

Chienchi

--
You received this message because you are subscribed to the Google Groups "edge-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to edge-users+...@googlegroups.com.
To post to this group, send email to edge-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/edge-users/fab8934d-202d-42dd-8852-822892bb39c4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Scott]

Hello Chienchi,

I just checked the run today. The assembly took 60 hrs to complete. In addition to the 5.4Mb bacterial genome, there were 4 plasmids 117Kb, 140Kb, 85Kb and 2Kb.  Depth of coverage ranged from 250x-400x. I did another assembly today with a different dataset using default settings.  Assembly only took 1 hr  07 min  (40x depth coverage). Thank you for the suggestion to turn off read correction.  I will try it when I have high depth of coverage.

Thanks,

Scott

On Monday, September 17, 2018 at 9:35:54 AM UTC-6, Lo, Chien-Chi wrote:

Hi Scott,

 

Is the run ever finished? We don’t have many nanopore use cases. Here are two cases we recently ran. The Ecoli (4.6 Mb) dataset with ~60x depth coverage took 1 hrs 06 mins using 8 CPUs.  Another 2Mb genome with 230x depth coverage took ~10 hours using 8 CPUs. The more depth coverage does affect the assembly time. In addition, by default, EDGE uses minimap2/racon to do reads correction before miniasm assembly in the lrasm pipeline.  We found the reads error correction step is usually the time bottleneck of the pipeline. You may try to turn the reads error correction off and see how it goes.

 

Thanks,

Chienchi

 

 

 

From: <edge-...@googlegroups.com> on behalf of Scott <starfi...@gmail.com>
Date: Friday, September 14, 2018 at 4:02 PM
To: edge-users <edge-...@googlegroups.com>
Subject: assembly times using lrasm

 

Hello,

I am using the lrasm pipeline to assemble a 5.4 Mb bacterial genome. According to contigs.log, minimap2 took 19 min to complete.  Racon was then started and has been running for 28 hrs using 7 threads.  Can anyone provide a general guideline for how long assembly should take (the computer running EDGE has 2 Intel Xeon E5-2620 v2 Ivy Bridge_EP 2.1 GHz processors (total of 12 cores) and 256GB ram)?  For comparison, the same data set was assembled using Unicycler which also used miniasm/minimap2 and racon.  Minasm took 15 min to assemble and racon polishing (9 rounds) took ~4.5 hrs using 8 threads.  The depth of coverage for the genome is ~250.

Thanks,

Scott 

--
You received this message because you are subscribed to the Google Groups "edge-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to edge-users+...@googlegroups.com.

To post to this group, send email to edge...@googlegroups.com.

[Chain, Patrick Sam Guy]

Thanks for the emails Scott – we’re implementing a few text information boxes to highlight the fact that if you are doing assembly, it’s best to forego read polishing, unless you’re using nanopolish or the like that uses native datafiles….

 

Best,

patrick

To post to this group, send email to edge-...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/edge-users/6d4df59f-eb6f-4cba-9c73-b7cfce02d6e4%40googlegroups.com.

[Scott]

I ran the assembly today without error correction.  Assembly time dropped from 60hrs to 1hr 44min.

Thanks,

Scott