Skip to main content

Table 2 k -mers counting results for Homo sapiens NA19238 individual (353 GB FASTQ file or 463 gzipped FASTQ files of total size 116.6 GB)

From: Disk-based k-mer counting on a PC

  k=22 k=25 k=28 k=31
Algorithm Space Time Space Time Space Time Space Time
  Classic counters
  32-core server
BFCounter 46/ 0 114,083 41/ 0 99,468 Failed Failed
Jellyfish 50/ 0 2,303 64/ 0 2,258 75/ 0 2,208 87/ 0 2,107
Jellyfish1 27/ 39 2,964 33/ 36 2,769 21/ 27 2,673 24/ 22 2,511
DSK 6/200 6,490 6/340 6,020 6/280 5,115 6/221 4,215
DSK gz 6/200 9,076 6/340 8,029 6/280 7,157 6/221 6,424
KMC 32/104 1,405 32/130 1,488 32/133 1,522 32/121 1,471
KMC 16/107 1,548 16/131 1,657 16/141 1,684 16/128 1,568
KMC gz 32/104 1,040 32/129 1,066 32/132 1,055 32/120 989
KMC gz 16/107 1,278 16/130 1,631 16/141 1,662 16/125 1,307
  6-core PC
DSK 6/200 21,468 6/340 18,774 6/280 15,384 6/221 11,857
DSK gz 6/200 18,939 6/340 16,818 6/280 14,694 6/221 12,070
KMC 11/107 3,482 11/128 3,442 11/138 3,584 11/127 3,515
KMC gz 11/107 2,198 11/128 2,206 11/138 2,303 11/127 2,365
  Quake-compatible counters
  32-core server
BFCounter 70/ 0 171,888 72/ 0 180,861 Failed Failed
Jellyfish 100/ 0 4,339 57/230 2,891 64/192 3,246 70/175 3,161
KMC 32/311 2,585 32/302 2,467 32/282 2,347 32/237 2,106
KMC 16/315 2,615 16/305 2,730 16/283 2,592 16/245 2,284
KMC gz 32/310 2,071 32/302 1,995 32/282 1,880 32/237 1,690
KMC gz 16/318 2,273 16/304 2,611 16/283 2,015 16/244 2,707
  6-core PC
KMC 11/316 5,538 11/298 5,533 11/277 5,184 11/242 5,016
KMC gz 11/316 5,370 11/298 5,060 11/277 4,708 11/243 4,643
  1. RAM and disk spaces (the first and the second value in the column “Space”, respectively) are in GB (1GB= 230B). Time is in seconds. The test machines: 32-core server, 6-core PC (see more details in the text). Superscripts denote: 1—RAM limited to 36GB, gz—input data in gzipped files. The programs were used for the number of threads adjusted to the number of cores to achieve maximum speed. The asterisk signs (for Jellyfish) denote that two separate databases were constructed by Jellyfish due to the memory limit of the machine (128GB RAM) and Jellyfish reported that to merge these databases it needs more RAM, so these times are underestimated.