R9.4 formatting issue

[yk-tanigawa]

• R9.4 is not fully supported by the following tools:
  • poretools
  • porekit

[yk-tanigawa]

One of the possible approach we can take is convert to fastq format and map to the reference (avoid handling nanopore data directly).

Or, we can spend time to understand how they changed their format (and fix tools).

[yk-tanigawa]

We decided to go for fastq approach 👍

[yk-tanigawa]

• I have observed the data structures inside two fast5 files:
  • one file from our cDNA data
  • the other from sample file in poretools

[yk-tanigawa]

• I used HDF5 view (https://support.hdfgroup.org/products/java/release/download.html#bin), which provides a graphical interface to observe hierarchy in hdf5 file (I can use it to fast5 file as well)

[yk-tanigawa]

Difference of data structure

FAST5 file from cDNA

FAST5 file what poretools expect to have

[yk-tanigawa]

what I found from this observation

• FASTQ file is stored in FAST5 file as string
  • poretools (and other software) just retrieve this information by using h5py package
• metadata is also stored in FAST5 file
  • poretools format:
    • Analysis/Basecall_1D_000/BaseCalled_template/Events etc.
  • cDNA format:
    • 'Analysis/Basecall_1D_000/Summary/basecall_1d_template
• Our FAST5 file still contains Raw signal
  • It might be interesting to see what kind of output file we will have after feeding this file to a base caller

[yk-tanigawa]

• I just created Base calling with Deepnano #10