Preface
When I trained a new model using bonito, remora or the others, the validation loss and accuracy are great indicators indeed. However, I am wandering what the single base and polyT(A) accuracy are like.
From Gimpel, A.L., Stark, W.J., Heckel, R. et al., it is known that errors are always biased in the entire DNA storage process. In addition, I am curious about the extent of this bias.
Basecalling
1 | bonito basecaller dna_r10.4.1_e8.2_400bps_hac@v5.0.0 ./pod5 > ./basecalls/sample_250410_e8.2_400bps_hac.fastq |
to be done …