One-hot encoding of biological sequence data in Go.
Many common neural network architectures can accept one-hot encoded data as input. GOHencode is a convenient tool to one-hot encode entire fasta files.
The input to GOHencode is a (gzipped) fasta file of sequences. These files have a structure as follows:
>Sequence-identifier
ATCGAA
GOHencode processes a fasta file and returns a parquet file with the following structure:
| Sequence ID | Encoded Sequence Array |
|---|---|
| Sequence identifier 1 | [4*len(sequence)] |
The encoded sequence array has the following form for the sample sequence above:
A | 1 | 0 | 0 | 0 | 1 | 1
C | 0 | 0 | 1 | 0 | 0 | 0
T | 0 | 1 | 0 | 0 | 0 | 0
G | 0 | 0 | 0 | 1 | 0 | 0
The one-hot encoded output is stored as a parquet file with the form:
| Base1 | Base2 | ... | Basen |
|---|---|---|---|
| [1 0 0 0] | [0 0 1 0] | ... | [...] |