Difference between revisions of "Applications/guppy"

From HPC
Jump to: navigation , search
m
m
Line 4: Line 4:
  
 
* Description: Local accelerated basecalling for Nanopore data.
 
* Description: Local accelerated basecalling for Nanopore data.
* Version: guppy 3.1.5
+
* Version: guppy 3.1.5 and 4.0.14
* Modules: guppy/cpu/3.1.5 (CPU) and guppy/gpu/3.1.5 (GPU)
+
* Modules: guppy/cpu/3.1.5 (CPU), guppy/gpu/3.1.5 (GPU), guppy/cpu/4.0.14 (CPU), guppy/gpu/4.0.14 (GPU)
 
* Licence: Free, open-source
 
* Licence: Free, open-source
  
Line 11: Line 11:
  
 
Using the GPU as accelerators will make the analysis considerably faster (on large data sets). For small runs the CPU version is recommended and probably faster by removing the GPU overhead.
 
Using the GPU as accelerators will make the analysis considerably faster (on large data sets). For small runs the CPU version is recommended and probably faster by removing the GPU overhead.
 +
 +
Important : guppy/gpu (GPU version) requires compute level of six or above, see examples below.
  
 
===Batch example===
 
===Batch example===
  
This uses 2 GPUs presently on GPU05 (compute level is high than 6) which is required by guppy's minimum requirement.
+
This uses 2 GPUs presently on GPU05 ('''compute level is high than 6''') which is required by guppy's minimum requirement.
  
 
<pre>
 
<pre>

Revision as of 11:22, 10 August 2020

Application Details

  • Description: Local accelerated basecalling for Nanopore data.
  • Version: guppy 3.1.5 and 4.0.14
  • Modules: guppy/cpu/3.1.5 (CPU), guppy/gpu/3.1.5 (GPU), guppy/cpu/4.0.14 (CPU), guppy/gpu/4.0.14 (GPU)
  • Licence: Free, open-source

Usage Examples

Using the GPU as accelerators will make the analysis considerably faster (on large data sets). For small runs the CPU version is recommended and probably faster by removing the GPU overhead.

Important : guppy/gpu (GPU version) requires compute level of six or above, see examples below.

Batch example

This uses 2 GPUs presently on GPU05 (compute level is high than 6) which is required by guppy's minimum requirement.

#!/bin/bash
#SBATCH -J guppy.job
#SBATCH --exclusive
#SBATCH -o gpu05.%j.out
#SBATCH -e gpu05.%j.err
#SBATCH -p gpu05
#SBATCH --gres=gpu:tesla:2
#SBATCH --exclusive


module purge
module load cuda/10.1.168
module load guppy/gpu/3.1.5


guppy_basecaller --input_path /home/user/fast5 --save_path /home/user/guppy --flowcell FLO-MIN106 --kit SQK-RPB004 --min_qscore 7 --qscore_filtering -x cuda:0 cuda:1

Compute capability

For applications that require this information:

  • K40m is 3.5
  • P100 is 6.0


Further Information

Navigation