Skip to content

Latest commit

 

History

History
40 lines (29 loc) · 3.71 KB

guppy5.md

File metadata and controls

40 lines (29 loc) · 3.71 KB
Raw

Guppy5 models

Jun 9, 2021. Guppy5 basecaller has been released recently. According to our observations, Guppy5 gives better base accuracy but has a substantially different error profile from the HAC mode in Guppy3 and Guppy4. While we are releasing Guppy5 models for Clair3 users to take advantage of the new Guppy5 basecaller, please aware that these models are not yet being heavily tested, and the Guppy5 data available for model training is still limited.

We fine-tuned the HG001,2,4,5 (Guppy3,4) model using ~70x HG002 data basecalled using the Guppy V5.0.7 using model dna_r9.4.1_450bps_sup from ONT (visited May 25th, 2021). The new model is incorporated into v0.1-r3 docker image with folder name ont_guppy5. The new model is also available here.

The benchmarks of the model and the data are as follows. With all precision increased slightly, the recall, espeicially the Indel recall, has improved significantly using Guppy5 data on a Guppy5 model (from 62.712% to 72.281%). The precision dropped when using Guppy5 data on the Guppy3,4 model, necessitating a Guppy5 model specially trained on Guppy5 data. The overall performance dropped signficiantly when using Guppy4 data on a Guppy5 model. We are not sure whether it's because the Guppy5 model was fine-tuned instead of trained from scratch, or it reflects the large differences between Guppy4 and Guppy5 data. A Guppy5 model trained from scratch will be provided when more Guppy5 data is available.

Model Testing
Dataset		 Overall
Precision		 Overall
Recall		 Overall
F1		 SNP
Precision		 SNP
Recall		 SNP
F1		 Indel
Precision		 Indel
Recall		 Indel
F1
Guppy5 model Guppy5 HG002 chr20 ~70x 98.440% 96.100% 97.256% 99.812% 99.858% 99.835% 88.056% 72.281% 79.393%
Guppy3,4 model Guppy5 HG002 chr20 ~70x 97.294% 95.502% 96.390% 99.717% 99.858% 99.788% 79.616% 67.893% 73.289%
Guppy3,4 model Guppy4 HG002 chr20 70x 98.481% 94.709% 96.558% 99.780% 99.758% 99.769% 87.278% 62.713% 72.984%
Guppy5 model Guppy4 HG002 chr20 70x 94.184% 93.809% 93.996% 99.535% 99.634% 99.585% 59.359% 56.894% 58.100%

Use the Guppy5 model

INPUT_DIR="[YOUR_INPUT_FOLDER]"           # e.g. /home/user1/input (absolute path needed)
OUTPUT_DIR="[YOUR_OUTPUT_FOLDER]"         # e.g. /home/user1/output (absolute path needed)
THREADS="[MAXIMUM_THREADS]"               # e.g. 8

docker run -it \
  -v ${INPUT_DIR}:${INPUT_DIR} \
  -v ${OUTPUT_DIR}:${OUTPUT_DIR} \
  hkubal/clair3::latest \
  /opt/bin/run_clair3.sh \
  --bam_fn=${INPUT_DIR}/input.bam \       ## change your bam file name here
  --ref_fn=${INPUT_DIR}/ref.fa \          ## change your reference name here
  --threads=${THREADS} \                  ## maximum threads to be used
  --platform="ont" \                       
  --model_path="/opt/models/r941_prom_sup_g506" \  
  --output=${OUTPUT_DIR}                  ## absolute output path prefix

Check Usage for more options.