Releases: apcamargo/RNAsamba
Releases · apcamargo/RNAsamba
0.2.5
- Fix a dependency conflict between
keras
and h5py
.
0.2.4
- Sequences shorter than 4 nucleotides would break the program in the k-mer frequency step (as RNAsamba uses 2, 3 and 4-mers). Now, those sequences are ignored.
0.2.3
- The input FASTA file is now read only once.
- Several unnecessary loops were removed.
0.2.2
- The
longest_orf
function is now written in Rust and is ~50 times faster.
- The ORF extraction step is now parallelized.
0.2.1
- The k-mer frequency function is now written in Rust and is parallelized.
0.2.0
- The
count_kmers
function is now written in Rust and is approximately 5.5× faster than the old native Python implementation.
0.1.6
- Class and function names in the
miniigloo
module were improved.
- Updated Keras version requirements.
0.1.5
- CLI now uses subparsers for the classification and training modules.
- Changed version requirements for numpy (<1.17) to avoid unnecessary warnings.
- Disabled Tensorflow's excessive logging and warnings.
- Step number is now shown.
0.1.4
- Replace
U
with T
in nucleotide sequences.
0.1.2
- Set the minimal Python version to 3.6
- Add the partial-length weights file
- Remove
tf.logging.set_verbosity(tf.logging.ERROR)
, which was causing warnings in TensorFlow 1.14