Limits...
FFBSKAT: fast family-based sequence kernel association test.

Svishcheva GR, Belonogova NM, Axenovich TI - PLoS ONE (2014)

Bottom Line: We compared the performance of our software with that of two existing software for family-based sequence kernel association testing, namely, ASKAT and famSKAT, using the Genetic Analysis Workshop 17 family sample.In addition, the calculations of the three-compared software were similarly accurate.With respect to the available analysis modes, we combined the advantages of both ASKAT and famSKAT and added new options to empower FFBSKAT users.

View Article: PubMed Central - PubMed

Affiliation: Institute of Cytology and Genetics, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia.

ABSTRACT
The kernel machine-based regression is an efficient approach to region-based association analysis aimed at identification of rare genetic variants. However, this method is computationally complex. The running time of kernel-based association analysis becomes especially long for samples with genetic (sub) structures, thus increasing the need to develop new and effective methods, algorithms, and software packages. We have developed a new R-package called fast family-based sequence kernel association test (FFBSKAT) for analysis of quantitative traits in samples of related individuals. This software implements a score-based variance component test to assess the association of a given set of single nucleotide polymorphisms with a continuous phenotype. We compared the performance of our software with that of two existing software for family-based sequence kernel association testing, namely, ASKAT and famSKAT, using the Genetic Analysis Workshop 17 family sample. Results demonstrate that FFBSKAT is several times faster than other available programs. In addition, the calculations of the three-compared software were similarly accurate. With respect to the available analysis modes, we combined the advantages of both ASKAT and famSKAT and added new options to empower FFBSKAT users. The FFBSKAT package is fast, user-friendly, and provides an easy-to-use method to perform whole-exome kernel machine-based regression association analysis of quantitative traits in samples of related individuals. The FFBSKAT package, along with its manual, is available for free download at http://mga.bionet.nsc.ru/soft/FFBSKAT/.

Show MeSH

Related in: MedlinePlus

Dependence of the running times of the second step of mini-exome analysis of quantitative trait Q1 on sample size for different methods (using one processor at 3.07 GHz).Points show the estimated running times (RT), lines correspond to the linear regression equations: RTASKAT = 9×10−6n3–.753; RTfamSKAT = 6.7×10–5n2–2.8, and RTFFBSKAT = 1.7×10–5n2–3.7, where n is the sample size.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4048315&req=5

pone-0099407-g001: Dependence of the running times of the second step of mini-exome analysis of quantitative trait Q1 on sample size for different methods (using one processor at 3.07 GHz).Points show the estimated running times (RT), lines correspond to the linear regression equations: RTASKAT = 9×10−6n3–.753; RTfamSKAT = 6.7×10–5n2–2.8, and RTFFBSKAT = 1.7×10–5n2–3.7, where n is the sample size.

Mentions: The results presented in Fig. 1 show that “ASKAT” displays cubic dependence on sample size, whereas “famSKAT” and “FFBSKAT” exhibit quadratic dependence. “FFBSKAT” is approximately four times faster than “famSKAT”. In our experiment, the “famSKAT” and “ASKAT” procedures were additionally optimized by excluding matrix operations whose operands were independent of genotypes operations. Therefore, the performance of the famSKAT and ASKAT software is expected to be considerably slow in practice.


FFBSKAT: fast family-based sequence kernel association test.

Svishcheva GR, Belonogova NM, Axenovich TI - PLoS ONE (2014)

Dependence of the running times of the second step of mini-exome analysis of quantitative trait Q1 on sample size for different methods (using one processor at 3.07 GHz).Points show the estimated running times (RT), lines correspond to the linear regression equations: RTASKAT = 9×10−6n3–.753; RTfamSKAT = 6.7×10–5n2–2.8, and RTFFBSKAT = 1.7×10–5n2–3.7, where n is the sample size.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4048315&req=5

pone-0099407-g001: Dependence of the running times of the second step of mini-exome analysis of quantitative trait Q1 on sample size for different methods (using one processor at 3.07 GHz).Points show the estimated running times (RT), lines correspond to the linear regression equations: RTASKAT = 9×10−6n3–.753; RTfamSKAT = 6.7×10–5n2–2.8, and RTFFBSKAT = 1.7×10–5n2–3.7, where n is the sample size.
Mentions: The results presented in Fig. 1 show that “ASKAT” displays cubic dependence on sample size, whereas “famSKAT” and “FFBSKAT” exhibit quadratic dependence. “FFBSKAT” is approximately four times faster than “famSKAT”. In our experiment, the “famSKAT” and “ASKAT” procedures were additionally optimized by excluding matrix operations whose operands were independent of genotypes operations. Therefore, the performance of the famSKAT and ASKAT software is expected to be considerably slow in practice.

Bottom Line: We compared the performance of our software with that of two existing software for family-based sequence kernel association testing, namely, ASKAT and famSKAT, using the Genetic Analysis Workshop 17 family sample.In addition, the calculations of the three-compared software were similarly accurate.With respect to the available analysis modes, we combined the advantages of both ASKAT and famSKAT and added new options to empower FFBSKAT users.

View Article: PubMed Central - PubMed

Affiliation: Institute of Cytology and Genetics, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia.

ABSTRACT
The kernel machine-based regression is an efficient approach to region-based association analysis aimed at identification of rare genetic variants. However, this method is computationally complex. The running time of kernel-based association analysis becomes especially long for samples with genetic (sub) structures, thus increasing the need to develop new and effective methods, algorithms, and software packages. We have developed a new R-package called fast family-based sequence kernel association test (FFBSKAT) for analysis of quantitative traits in samples of related individuals. This software implements a score-based variance component test to assess the association of a given set of single nucleotide polymorphisms with a continuous phenotype. We compared the performance of our software with that of two existing software for family-based sequence kernel association testing, namely, ASKAT and famSKAT, using the Genetic Analysis Workshop 17 family sample. Results demonstrate that FFBSKAT is several times faster than other available programs. In addition, the calculations of the three-compared software were similarly accurate. With respect to the available analysis modes, we combined the advantages of both ASKAT and famSKAT and added new options to empower FFBSKAT users. The FFBSKAT package is fast, user-friendly, and provides an easy-to-use method to perform whole-exome kernel machine-based regression association analysis of quantitative traits in samples of related individuals. The FFBSKAT package, along with its manual, is available for free download at http://mga.bionet.nsc.ru/soft/FFBSKAT/.

Show MeSH
Related in: MedlinePlus