search

jts / nanopolish

Signal-level algorithms for MinION data
MIT License
550 stars 160 forks source link

Nanopolish output: no alignments #1095

Open tait123 opened 11 months ago

tait123 commented 11 months ago

Hello, I’m working on running nanopolish for methylation calling. My command line was: nanopolish call-methylation -t 8 -r /home/choilab_tk/nanopore/data/221108_bank7492_NIID_Cas9/20221108_0927_X5_FAV43835_7491c0d0/fastq_pass/pass/merged/output_merged.fastq -b /home/choilab_tk/nanopore/results/221108_bank7492/minimap2/merged/output_merged.bam -g /home/choilab_tk/hg38/hg38.fa -w "chr1:149,390,100-149,391,900" > methylation_calls.tsv

However, when I run above command line, output shows: [bam process] iterating over region: chr1:149,390,100-149,391,900 [post-run summary] total reads: 26, unparseable: 0, qc fail: 0, could not calibrate: 0, no alignment: 13, bad fast5: 0

Any recommendation on why 50% of total reads cannot be aligned? and how to solve this problem?

For you information, I used Guppy for basecalling, and it command line was: guppy_basecaller -i /home/choilab_tk/nanopore/data/221108_bank7492_NIID_Cas9/20221108_0927_X5_FAV43835_7491c0d0/fast5_pass -s /home/choilab_tk/nanopore/data/221108_bank7492_NIID_Cas9/20221108_0927_X5_FAV43835_7491c0d0/fastq_pass -c dna_r10.4.1_e8.2_260bps_fast.cfg --num_callers 4 --trim_adapters --trim_primers --compress_fastq

jts commented 11 months ago

It looks like your data is R10, please see here: https://github.com/jts/nanopolish#a-note-on-r10-support

tait123 commented 11 months ago

Thank you for the reply.

According to the link you provided, it recommended to use evenalign command. Also, I noticed that event_level_mean histogram from methylation detection paper demonstrate how methylation chenges the observed current.

I wonder if there is any tutorial for plotting the histogram using ggplot2.

jts commented 11 months ago

Sorry but we do not support R10 for nanopolish eventalign yet - you may wish to look at f5c as it does support R10 data (cc: @hasindu2008)

hasindu2008 commented 11 months ago

@tait123

f5c is here https://github.com/hasindu2008/f5c/ and the interface is similar to nanopolish. It supports eventalign and call-methylation for R10. For using f5c with R10, specify --pore r10. If you need some help with it, you can open an issue in f5c repository.