Data preparation from raw reads

Raw data from Oxford Nanopore Technologies sequencing

With the introduction of R10.4.1 flow cell, the diversity of ONT data formats has increased. These now include the original single and multiple fast5 format, the newer pod5 format, and the community-driven slow5/blow5 formats. The relationship between conversion tools and these different data formats is as below.

nanoCEM only supports single-format fast5 and blow5; please transfer the data format before usage. Since we support tombo, move_table and f5c, while tombo only supports single format fast5 and our f5c and move_table mode supports blow5. Here are some advice commands if your original data is multi-format fast5.

# pod5 to blow5
blue-crab p2s file.pod5 -o file.blow5

# fasts to blow5
slow5tools f2s </path/to/multi_reads> -d </path/to/blow5_dir>
slow5tools cat </path/to/blow5_dir> -o file.blow5

Basecall your raw reads

After obtaining raw reads files, the first step is to basecall them. Here is an example script to run Guppy and Dorado basecaller. You can find more details about basecalling at ONT:

# Guppy basecaller
guppy_basecaller -i <path/to/fastt> -s <path/to/fastq> --config <config file> --device auto -r
cat <path/to/fastq> > file.fastq
# Dorado basecaller (support fast5 & pod5)
dorado basecaller <model> </path/to/reads> > dorado.bam
samtools bam2fq dorado.bam > file.fastq

Choose your reference

The alignment of DNA is relatively simple, but for RNA, it becomes more complex due to the presence of alternative splicing and multiple isoforms in eukaryotic organisms.

The alignment process is already embedded in nanoCEM. For general analysis, if you use RNA mode, it is advisable to use transcriptome as the reference, while for DNA, the reference would be the genome.

This aligns signal with sequence via a table of expected signal levels. After a signal mapping has been refined it becomes more comparable to other reads enabling powerful downstream analyses. nanoCEM supports the re-squiggle results of f5c resquiggle, f5c eventalign, move_table and tombo. while f5c has already been integrated into our script, tombo users will need to install the required environment on their own env and move_table requires the additional option in basecaller to output more tags in the bam file. Due to Tombo being an older tool and no longer receiving active updates, it presents challenges for seamless integration into our pipeline.

Tombo

Optional: We provide a python script extract_sub_fast5_from_bam to assist in sampling reads aligned to your region of interest to reduce re-squiggle time.

extract_sub_fast5_from_bam -i </path/to/single_reads> -o </path/to/results> -b file.bam --chrom NR_103073.1 --pos 2030 --cpu 32

Tombo is a suite of tools primarily for the identification of modified nucleotides from nanopore sequencing data, and firstly proposed the generation of re-squiggle.

tombo preprocess annotate_raw_with_fastqs --fast5-basedir  </path/to/single_reads> --fastq-filenames file.fastq --processes 16 
tombo resquiggle </path/to/single_reads> reference.fasta --processes 16 --num-most-common-errors 5

Notes: Since the re-squiggle process in tombo can be time-consuming, we recommend using an SSD (Solid State Drive) to reduce processing time.

move_table

The move table records the index of events in basecalling. In the resulting bam file from basecalling, this information is stored in the tags ns, ts and mv. You can obtain this information using the following basecalling commands.

Notes: For dorado, recommend to use Dorado 0.4.3 to generate basecalled bam files with move_table, due to differences between event index length in bam and signal length in pod5 from Dorado 0.5.2. This issue has been reported on their GitHub and expected to address in the following updates.