Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Extended Islands of Tractability for

Parsimony Haplotyping

Rudolf Fleischer1 , Jiong Guo2 , Rolf Niedermeier3 , Johannes

Uhlmann3 , Yihui Wang1 , Mathias Weller3 , and Xi Wu1

1 Fudan University Shanghai, 2 Universität des Saarlandes,

and 3 Friedrich-Schiller-Universität Jena

January 7, 2015

1 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

First, Some Biology...

A G T T A G C G A

A G T C A G C A A

gene gene
approx. 0.1% of human nucleotide sites differ between individuals

2 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

First, Some Biology...

A G T T A G C G A

A G T C A G C A A

gene gene
approx. 0.1% of human nucleotide sites differ between individuals

2 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

First, Some Biology...

A G T T A G C G A

A G T C A G C A A

gene gene
the sequence of SNPs is called a haplotype

2 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

First, Some Biology...

T G

C A

the sequence of SNPs is called a haplotype

2 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

First, Some Biology...

humans are diploid ; 2 chromosome sets

haplotype = SNP sequence in one chromosome set
genotype = SNP sequence in the combined chromosome sets

3 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

First, Some Biology...

humans are diploid ; 2 chromosome sets

haplotype = SNP sequence in one chromosome set
genotype = SNP sequence in the combined chromosome sets

3 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Motivation

Haplotype Inference
goal: find relation between certain SNPs and genetic diseases
problem: difficult (expensive) to sequence both haplotypes
but: easy (cheap) to sequence the genotype instead
; idea: sequence genotype and computationally infer haplotypes

4 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Motivation

Haplotype Inference
goal: find relation between certain SNPs and genetic diseases
problem: difficult (expensive) to sequence both haplotypes
but: easy (cheap) to sequence the genotype instead
; idea: sequence genotype and computationally infer haplotypes

Problems
impossible to infer haplotypes of just 1 genotype
; sequence and infer groups/populations
which explanation should be preferred if there are multiple?
; parsimony
how to perform the actual computation fast?

4 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Parsimony

Parsimony (Ockham’s razor)

Under many plausible explanations of an observed phenomenon,
the one requiring the fewest assumptions should be preferred.

5 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Parsimony

Parsimony (Ockham’s razor)

Under many plausible explanations of an observed phenomenon,
the one requiring the fewest assumptions should be preferred.

used in...
... Clark’s problem [Clark, Molecular Biology and Evolution ’90]

... pure parsimony haplotyping [Lancia et al., INFORMS ’04]

... minimum perfect phylogeny [Gusfield & Orzack, HBI ’05]

... k-minimum recombination configuration [Li & Jiang, JBCB ’03]

...

5 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Preliminary Definitions

Definition (Haplotype, Genotype, Resolve)

haplotype h = string over {0, 1}
genotype g = string over {0, 1, 2}
h1 , h2 resolve g ⇔
for all i ∈ N, g [i] = h1 [i] = h2 [i] or g [i] = 2 and h1 [i] 6= h2 [i]
multiset H ; res (H)
multiset H resolves multiset G ⇔ G ⊆ res (H)

Example
haplotype1: 0 0 1 0 1 1 1 0 0 0 1
haplotype2: 0 0 1 1 0 1 0 0 1 1 1
genotype: 0 0 1 2 2 1 2 0 2 2 1

6 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Preliminary Definitions

Definition (Haplotype, Genotype, Resolve)

haplotype h = string over {0, 1}
genotype g = string over {0, 1, 2}
h1 , h2 resolve g ⇔
for all i ∈ N, g [i] = h1 [i] = h2 [i] or g [i] = 2 and h1 [i] 6= h2 [i]
multiset H ; res (H)
multiset H resolves multiset G ⇔ G ⊆ res (H)

Example
haplotype1: 0 0 1 0 1 1 1 0 0 0 1
haplotype2: 0 0 1 1 0 1 0 0 1 1 1
genotype: 0 0 1 2 2 1 2 0 2 2 1

6 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Preliminary Definitions

Definition (Haplotype, Genotype, Resolve)

haplotype h = string over {0, 1}
genotype g = string over {0, 1, 2}
h1 , h2 resolve g ⇔
for all i ∈ N, g [i] = h1 [i] = h2 [i] or g [i] = 2 and h1 [i] 6= h2 [i]
multiset H ; res (H)
multiset H resolves multiset G ⇔ G ⊆ res (H)

Example
haplotype1: 0 0 1 1 1 1 1 0 0 0 1
haplotype2: 0 0 1 0 0 1 0 0 1 1 1
genotype: 0 0 1 2 2 1 2 0 2 2 1

6 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Preliminary Definitions

Definition (Haplotype Graph)

Given H and G ⊆ res (H). haplotype graph of H and G :
|H| vertices (labeled by H)
|G | edges (labeled by G )
haplotypes of each edge resolve its genotype

Example
01001 11111
21221
1
11011 2102
22222

11122
12
21
2

12120
10110 11100

7 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Preliminary Definitions

Definition (Haplotype Graph)

Given H and G ⊆ res (H). haplotype graph of H and G :
|H| vertices (labeled by H)
|G | edges (labeled by G )
haplotypes of each edge resolve its genotype

Example
01001 11111
21221
1
11011 2102
22222

11122
12
21
2

12022
10110 11100

7 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Haplotype Inference by Parsimony

Definition (Haplotype Inference by Parsimony)

Input: multiset G of length-m genotypes, integer k ≥ 0
Question: ∃ multiset H of k haplotypes that resolves G ?

Example
genotypes haplotype graph haplotypes
11122 01001
21221 11111 01001
12120 11011 2102
1
10110
12212
22222

11122
11011
21021
12

11100
21
2

21221 12120 11111

22222 10110 11100

8 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Previous Work

Complexity
NP-hard [Halldórsson et al., DMTCS ’03]

APX-hard [Lancia et al., INFORMS ’04]

many special cases in P (e.g. [Lancia & Rizzi, ORL ’06])

Algorithms
O(2|G |·d ) Branch&Bound [Wang & Xu, Bioinformatics ’03]

ILP [Lancia & Serafini, INFORMS ’08]

2
O(mk 2k ) FPT algorithm [Sharan et al., TCBB ’06]

Factor-2d−1 -Approximation [Lancia & Rizzi, ORL ’06]

Constrained Version [Fellows et al., CPM ’09]

k := #haplotypes, m := stringlength, d := max #of 2’s in a genotype

9 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Our Contribution

polynomial-time solvable special case:

“Induced Haplotype Inference”
k 4k · poly (|G |, m) time algorithm
simple O(2k · k 2 )-bit problem kernel

10 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Induced Haplotype Inference

remember: H resolves G ⇔ G ⊆ res (H)

what if G = res (H)? ; haplotype graph is a clique
Definition
G = res (H) ; H induces G

11 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Induced Haplotype Inference

remember: H resolves G ⇔ G ⊆ res (H)

what if G = res (H)? ; haplotype graph is a clique
Definition
G = res (H) ; H induces G

Definition (Induced Haplotype Inference by Parsimony)

Input: multiset G of length-m genotypes
Question: ∃ multiset H of haplotypes that induces G ?

11 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Example
2 0 1 2 1
2 0 2 1 1
2 2 1 1 2 10 genotypes in total ; 5 haplotypes
2 2 1 2 2
1 0 2 2 1
1 2 1 0 2
1 2 1 2 2
1 2 2 1 2
1 2 2 2 2
1 1 1 2 0

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Example
2 0 1 2 1
2 0 2 1 1
2 2 1 1 2 10 genotypes in total ; 5 haplotypes
2 2 1 2 2 6 genotypes with 1 ; 4 haplotypes with 1
1 0 2 2 1 0 genotypes with 0 ; 1 haplotype with 0
1 2 1 0 2
1 2 1 2 2
1 2 2 1 2
1 2 2 2 2
1 1 1 2 0

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Example
2 0 1 2 1
2 0 2 1 1
2 2 1 1 2 10 genotypes in total ; 5 haplotypes
2 2 1 2 2 6 genotypes with 1 ; 4 haplotypes with 1
1 0 2 2 1 0 genotypes with 0 ; 1 haplotype with 0
1 2 1 0 2 6 genotypes in total ; 4 haplotypes
1 2 1 2 2
1 2 2 1 2
1 2 2 2 2
1 1 1 2 0

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Example
2 0 1 2 1
2 0 2 1 1
2 2 1 1 2 10 genotypes in total ; 5 haplotypes
2 2 1 2 2 6 genotypes with 1 ; 4 haplotypes with 1
1 0 2 2 1 0 genotypes with 0 ; 1 haplotype with 0
1 2 1 0 2 6 genotypes in total ; 4 haplotypes
1 2 1 2 2 1 genotype with 1 ; 2 haplotypes with 1
1 2 2 1 2 1 genotype with 0 ; 2 haplotypes with 0
1 2 2 2 2
1 1 1 2 0

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Example
2 0 1 2 1
2 0 2 1 1
2 2 1 1 2 10 genotypes in total ; 5 haplotypes
2 2 1 2 2 6 genotypes with 1 ; 4 haplotypes with 1
1 0 2 2 1 0 genotypes with 0 ; 1 haplotype with 0
1 2 1 0 2 6 genotypes in total ; 4 haplotypes
1 2 1 2 2
1 2 2 1 2
1 2 2 2 2
1 1 1 2 0

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Observation
G2 6= ∅ but G0 = ∅ or G1 = ∅ ; poly
|G0 | = |G1 | = 1 ; poly (although we may get 2 solutions)

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Observation
G2 6= ∅ but G0 = ∅ or G1 = ∅ ; poly
|G0 | = |G1 | = 1 ; poly (although we may get 2 solutions)

Strategy: Divide & Conquer

divide-step
base cases
merge-step

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Observation
G2 6= ∅ but G0 = ∅ or G1 = ∅ ; poly
|G0 | = |G1 | = 1 ; poly (although we may get 2 solutions)

Strategy: Divide & Conquer

divide-step ; OK
base cases
merge-step

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Observation
G2 6= ∅ but G0 = ∅ or G1 = ∅ ; poly
|G0 | = |G1 | = 1 ; poly (although we may get 2 solutions)

Strategy: Divide & Conquer

divide-step ; OK
base cases ; OK
merge-step

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations & Algorithm

Observation
G can be “nicely” partitioned into G0 , G1 , and G2 .

Observation
G2 6= ∅ but G0 = ∅ or G1 = ∅ ; poly
|G0 | = |G1 | = 1 ; poly (although we may get 2 solutions)

Strategy: Divide & Conquer

divide-step ; OK
base cases ; OK
merge-step ; problem!
need to find a way to compute H1 for given H0 and G2

12 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Extending a Subclique Solution

Observation
Let H0 induce G0 and let g be a genotype in G2 with the smallest
number of 2’s. ; All h ∈ H0 that are consistent with g are
equal.

13 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Extending a Subclique Solution

Observation
Let H0 induce G0 and let g be a genotype in G2 with the smallest
number of 2’s. ; All h ∈ H0 that are consistent with g are
equal.

Proof Idea
H1 H0

h’=1100... g=2120...
h=0110...

13 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Extending a Subclique Solution

Observation
Let H0 induce G0 and let g be a genotype in G2 with the smallest
number of 2’s. ; All h ∈ H0 that are consistent with g are
equal.

Proof Idea
H1 H0

h’=1100... g=2120...
h=0110...

g’

h’’=01?0...

13 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Concluding The Induced Problem

Divide & Conquer like algorithm

divide step O(|G |)
base solutions O(m)
extend (merge) O(|G2 | · |Hx | · m)
all in all O(|G | · k · m)

14 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Concluding The Induced Problem

Divide & Conquer like algorithm

divide step O(|G |)
base solutions O(m)
extend (merge) O(|G2 | · |Hx | · m)
all in all O(|G | · k · m)

Theorem
Induced Haplotype Inference by Parsimony can be solved in
O(|G | · k · m) time.

14 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

General Haplotype Inference by Parsimony

now: drop the clique-constraint

; arbitrary haplotype graph
; NP-hard

15 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

General Haplotype Inference by Parsimony

now: drop the clique-constraint

; arbitrary haplotype graph
; NP-hard
Idea
If we knew only the genotype labels of the correct haplotype graph,
could we reconstruct the correct haplotype labels?
; inference graph

15 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Definitions

Definition (Inference Graph)

inference graph Γ of G = an order-k graph with edges
consistently labeled by the genotypes in G

Example
21221
1
2102
22222

11122
12
21
2

12120

16 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations

Observation (non-bipartite components of Γ)

C = cycle in Γ
; for all i, |{g ∈ C | g [i] = 2}| is even
non-bipartite components of Γ ; O(|Γ| · m) time

Example
21221
1
2102
22222

11122
12
21
2

12120

17 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations

Observation (bipartite components of Γ)

all g [i] = 2 for some i ⇒ choose arbitrarily
bipartite components of Γ ; O(|Γ| · m) time

Example
21221
22222

11122
12120

18 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations

Observation
(G , k) yes-instance with solution H
⇒ ∃ Γ extendable (O(|Γ| · m) time) to a haplotype graph of H and G

19 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations

Observation
(G , k) yes-instance with solution H
⇒ ∃ Γ extendable (O(|Γ| · m) time) to a haplotype graph of H and G
algorithmic idea: guess Γ

19 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Observations

Observation
(G , k) yes-instance with solution H
⇒ ∃ Γ extendable (O(|Γ| · m) time) to a haplotype graph of H and G
algorithmic idea: guess Γ
better idea: guess a “spanning” subgraph of Γ

19 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Solving The Problem

Algorithm
1 guess “spanning” subgraph of Γ
2 infer the haplotype multiset H ; O(k · m) time
3 check whether H resolves G ; O(k 2 · m) time

20 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Solving The Problem

Algorithm
1 guess “spanning” subgraph of Γ
1 guess a size-k genotype subset of G ; O(k 2k ) possibilities
2 for these genotypes, guess 2 (of k) vertices ; O(k 2k ) possibilities
2 infer the haplotype multiset H ; O(k · m) time
3 check whether H resolves G ; O(k 2 · m) time

20 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Solving The Problem

Algorithm
1 guess “spanning” subgraph of Γ
1 guess a size-k genotype subset of G ; O(k 2k ) possibilities
2 for these genotypes, guess 2 (of k) vertices ; O(k 2k ) possibilities
2 infer the haplotype multiset H ; O(k · m) time
3 check whether H resolves G ; O(k 2 · m) time

Theorem
Haplotype Inference by Parsimony can be solved in O(k 4k+2 · m)
time.

20 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Example

11101 01011
21221

12222

01022
22020
10010 01000
01001 11111
21221
1
11011 2102
22222

11122
12
21
2

12120
10110 11100

21 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Example

21221

12222

01022
22020

21221
1
2102
22222

11122
12
21
2

21 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Example

21221

12222

01022
22020

?10?1
21221
1
2102
22222

11122
12
21
2

21 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Example

11??1 010?1
21221

12222

01022
22020
1?0?0 010?0
?10?1 111?1
21221
1
11011 2102
22222

11122
12
21
2

1??1? 111??

21 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Example

111?1 010?1
21221

12222

01022
22020
100?0 010?0
01001 11111
21221
1
11011 2102
22222

11122
12
21
2

10110 11100

21 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Example

11101 01011
21221

12222

01022
22020
10010 01000
01001 11111
21221
1
11011 2102
22222

11122
12
21
2

10110 11100

21 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Kernelization: A Simple Exponential Kernel

In the following: Matrix Representation MG and MH

22 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Kernelization: A Simple Exponential Kernel

In the following: Matrix Representation MG and MH

Observation
Equal columns in MG ; no new constraints or information

22 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Kernelization: A Simple Exponential Kernel

In the following: Matrix Representation MG and MH

Observation
Equal columns in MG ; no new constraints or information

Idea: remove all but one of these columns

22 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Kernelization: A Simple Exponential Kernel

In the following: Matrix Representation MG and MH

Observation
Equal columns in MG ; no new constraints or information

Idea: remove all but one of these columns

Reduction Rule
For all equal columns i,j of MG , delete column j.

22 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Correctness, Size, and Example

Lemma
MH resolves MG :
columns i,j of MH equal ⇒ columns i,j of MG equal

23 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Correctness, Size, and Example

Lemma
MH resolves MG :
columns i,j of MH equal ⇒ columns i,j of MG equal

Example
MG MH
1 1 1 2 1 2
0 1 0 0 0 1
1 2 1 2 1 0
1 0 1 1 1 0
1 2 2 1 2 2
1 1 0 1 0 1
2 1 0 2 0 1
1 1 1 0 1 1
2 1 2 2 2 1
1 1 1 1 1 1
2 2 2 2 2 2

23 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Correctness, Size, and Example

Lemma
MH resolves MG :
columns i,j of MH equal ⇒ columns i,j of MG equal

Example
MG MH
1 1 1 2 1 2
0 1 0 0 0 1
1 2 1 2 1 0
1 0 1 1 1 0
1 2 2 1 2 2
1 1 0 1 0 1
2 1 0 2 0 1
1 1 1 0 1 1
2 1 2 2 2 1
1 1 1 1 1 1
2 2 2 2 2 2

23 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Correctness, Size, and Example

Lemma
MH resolves MG :
columns i,j of MH equal ⇒ columns i,j of MG equal

Example
MG MH
1 1 1 2 2
0 1 0 0 1
1 2 1 2 0
1 0 1 1 0
1 2 2 1 2
1 1 0 1 1
2 1 0 2 1
1 1 1 0 1
2 1 2 2 1
1 1 1 1 1
2 2 2 2 2

23 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Correctness, Size, and Example

Lemma
MH resolves MG :
columns i,j of MH equal ⇒ columns i,j of MG equal

Example
MG MH
1 1 1 2 1 2
0 1 0 0 0 1
1 2 1 2 1 0
1 0 1 1 1 0
1 2 2 1 2 2
1 1 0 1 0 1
2 1 0 2 0 1
1 1 1 0 1 1
2 1 2 2 2 1
1 1 1 1 1 1
2 2 2 2 2 2

23 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Kernel Conclusion

Theorem
columns: ≤ 2k , rows: ≤ k2

; overall kernel size: 2k · k2

computation in O(|G | · m log m) time

; previous algorithm: O(k 4k+2 · 2k + k 2 · m log m) time

24 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Conclusion

what we saw. . .
introduced induced variant with O(k 3 · m) time algorithm
2
improved 2O(k log k) time algorithm to 2O(k log k) time
presented O(2k · k 2 )-bit kernel

25 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Conclusion

what we saw. . .
introduced induced variant with O(k 3 · m) time algorithm
2
improved 2O(k log k) time algorithm to 2O(k log k) time
presented O(2k · k 2 )-bit kernel

also in the paper

results also hold for sets instead of multisets
algorithmic results basically also hold for constrained variant

25 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Conclusion

what we saw. . .
introduced induced variant with O(k 3 · m) time algorithm
2
improved 2O(k log k) time algorithm to 2O(k log k) time
presented O(2k · k 2 )-bit kernel

also in the paper

results also hold for sets instead of multisets
algorithmic results basically also hold for constrained variant

future work
find polynomial kernel (or prove nonexistence)
distance from triviality measures
find 2O(k) time algorithm
25 / 26
Introduction Induced Haplotyping Improved FPT algorithm Simple Kernel Conclusion

Thank you

26 / 26