Skip to main content

Table 10 Lerna and Athena run on 7 Illumina short read datasets

From: Lerna: transformer architectures for configuring error correction tools for short- and long-read genome sequencing

Dataset Coverage Genome Read length (bp) #Reads Athena Lerna Speedup
D1 80\(\times\) E. coli str. K-12 substr 136 20.8M 98s 5.5s 17.8\(\times\)
D2 71\(\times\) E. coli str. K-12 substr 47 7.1M 49s 3s 16.3\(\times\)
D3 173\(\times\) Acinetobacter sp. ADP1 36 18.1M 69s 4s 17.25\(\times\)
D4 62\(\times\) B. subtilis 75 3.5M 52s 3s 17.3\(\times\)
D5 166\(\times\) L. interrogans C sp. ADP1 100 7.1M 100s 5.5s 18.2\(\times\)
D6 70\(\times\) A. thaliana 250 33.6M 400s 23s 17.4\(\times\)
D7 67\(\times\) Homo sapiens 101 202M 960s 63s 15.2\(\times\)
  1. The time for calculating perplexity has been reported, along with the read lengths and number of reads. We observe that on average our pipeline is 18\(\times\) faster than Athena. This translates to 80\(\times\) to 275\(\times\) faster than estimating the alignment rate with Bowtie2