bsaintjo / pod5-rs Goto Github PK

Hello @bsaintjo,

Thank you very much for working on this project! I found the repo some few weeks ago and it seemed inactive so I tried to pick up the work from where it was at the time. I wanted to get something working and maybe open a pull request with it. I don't know Rust at all (I'm using this project as a vehicle for learning it) so I've been making quite a slow progress.

I just managed to start reading the signal data and was observing that it seems that I'm getting different results than the ONT's pod5 Python library. And seeing that you've started working on the project again, I thought it might be a good idea to report it. It's quite possible that there's no problem at all and it's just my ignorance, so apologies if that's the case.

Steps to reproduce:

I tried to look at the raw compressed signal data for the first read in file extra/multi_fast5_zip_v3.pod5 (that is read 0000173c-bf67-44e7-9a9c-1ad0bc728e74).

I used my implementation to get the compressed signal and compared the first few bytes to ones I retrieve with ONT's Python library in the following manner:

import pod5
import numpy as np

reader =  pod5.Reader("extra/multi_fast5_zip_v3.pod5")
read = next(reader.reads())
batch, batch_index, row_index = read._find_signal_row_index(0)
read_compressed_signal  = batch.signal[row_index].as_py()  # get the signal row as bytes
signal_row_samples = batch.samples[row_index].as_py() # get the number of samples in the signal row
compressed_signal_array = np.array(list(read_compressed_signal))
# The bytes in compressed_signal_array are the same as the ones I get for the compressed signal from my rust implementation
decompressed_signal = pod5.vbz_decompress_signal(compressed_signal_array, signal_row_samples)
# The bytes in decompressed_signal are different than the ones I get from using your implementation of svb16.

And the way I decompress the signal in Rust:

let mut decoded = decode(compressed_signal, samples).unwrap();

where both compressed_signal and samples have the same values as the ones passed to the Python decompression function, but the result from the decompression is different.

PS: You can use the python snippet and compare the decompressed_signal bytes from it with the test function in reader.rs.