Algorithm

Complexity

Running time



Time

Space

Drosophila

Human


OPTIMA

O((m−c) δ^{3}
#seeds)

O((m−c)^{2}+c
n)

54 m

36 days

Gentig v.2 (d)

O(#it
m
δ^{3}
#hashes)

O(m^{2}+n+HashTable)

1.32 h

75 days

Gentig v.2 (tp)
  
1.85 h

174 days

SOMA v.2 (v)

O(m^{2}
n^{2})

O(m
n)

1.28 years

1,067 years

Likelihood (d+a)

O(m
n
δ^{2})

O(m
n)

22.22 h

2.72 years

Likelihood (d+a+t)
  
19.62 h

2.38 years

Likelihood (p+a+t)
  
41.73 h

5.53 years

 Running times reported are estimated from 2100 maps and extrapolated for the full datasets (82,000 Drosophila maps and 2.1 million human maps, for 100 × coverage; singlecore computation on Intel x86 64bit Linux workstations with 16GB RAM). The best columnwise running times are reported in bold. Note that including the permutationbased statistical tests for SOMA and the likelihood method would increase their runtime by a factor of greater than 100. The complexity analysis refers to maptosequence glocal alignment per map, where n is the total length of the in silico maps (\(\thicksim \)500,000 fragments for the human genome), m≪n is the length of the experimental map in fragments (typically 17 fragments on average), #seeds, c (default of two) and δ are as defined in the “Methods” section and #it (number of iterations), #hashes (geometric hashes found to match) and HashTable are as specified in [17, 24]