A multifactorial evaluation framework for gene regulatory network reconstruction

Atte Aalto; Johan Markdahl; Jorge Goncalves; Laurent Mombaerts

arxiv: 1906.12243 · v1 · pith:ADA2LNYXnew · submitted 2019-06-28 · 🧬 q-bio.MN · math.DS· q-bio.QM

A multifactorial evaluation framework for gene regulatory network reconstruction

Laurent Mombaerts , Atte Aalto , Johan Markdahl , Jorge Goncalves This is my paper

Pith reviewed 2026-05-25 13:29 UTC · model grok-4.3

classification 🧬 q-bio.MN math.DSq-bio.QM

keywords gene regulatory networksnetwork inferencetime-series dataperturbationsrhythmic systemsbenchmark modelsexperimental designgene knock-out

0 comments

The pith

For rhythmic gene regulatory systems, long time-series improve network inference more than additional perturbations, while the reverse holds for non-rhythmic systems.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests how gene network reconstruction algorithms respond to changes in the amount of time-series data and the number of perturbations applied to the system. It generates data from six standard benchmark models that represent both rhythmic and non-rhythmic dynamics, then feeds increasing lengths of these series to five established inference methods. The evaluation reveals that the methods gain unequally from extra data and that the best allocation of experimental effort depends on system type. Rhythmic cases profit more from extending series length than from adding separate perturbation experiments, while non-rhythmic cases show the opposite pattern. These patterns supply concrete rules for deciding whether to lengthen recordings or run more conditions when resources are limited.

Core claim

The algorithms do not benefit equally from data increments. For rhythmic systems, it is more profitable for network inference strategies to be run on long time-series rather than short time-series with multiple perturbations. By contrast, for the non-rhythmic systems, increasing the number of perturbation experiments yielded better results than increasing the sampling frequency.

What carries the argument

A multifactorial evaluation framework that generates realistic time-series data from six benchmark models under controlled increases in series length and perturbation count, then measures reconstruction accuracy across five distinct inference algorithms.

If this is right

Benchmark studies of network inference should routinely vary both series length and perturbation number instead of fixing one factor.
Experimental budgets for rhythmic systems are better spent on longer continuous recordings than on additional separate perturbations.
Experimental budgets for non-rhythmic systems are better spent on more distinct perturbation conditions than on denser sampling within fewer conditions.
Algorithm design can incorporate knowledge of system rhythmicity to weight the value of additional time points versus additional experiments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The rhythmic versus non-rhythmic distinction may reflect whether repeated cycles or diverse steady-state conditions are needed to expose regulatory edges.
Similar multifactorial testing could be extended to other inference tasks such as signaling pathway reconstruction or metabolic network inference.
The framework suggests that hybrid experimental strategies, mixing moderate length with moderate perturbations, may be suboptimal for both system classes.

Load-bearing premise

The six widely used benchmark models together with the random perturbations applied to them produce data whose statistical properties match those of real experimental gene-regulatory time series sufficiently well for the performance rankings to generalize.

What would settle it

Apply the same five algorithms to real experimental time-series data from known rhythmic and non-rhythmic gene networks and observe that the performance advantage of long series versus multiple perturbations reverses or vanishes.

Figures

Figures reproduced from arXiv: 1906.12243 by Atte Aalto, Johan Markdahl, Jorge Goncalves, Laurent Mombaerts.

**Figure 2.** Figure 2: Evaluation of the effects of dynamical transients [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Figure 3: Area Under the ROC curve and the Precision [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

read the original abstract

In the past years, many computational methods have been developed to infer the structure of gene regulatory networks from time-series data. However, the applicability and accuracy presumptions of such algorithms remain unclear due to experimental heterogeneity. This paper assesses the performance of recent and successful network inference strategies under a novel, multifactorial evaluation framework in order to highlight pragmatic tradeoffs in experimental design. The effects of data quantity and systems perturbations are addressed, thereby formulating guidelines for efficient resource management. Realistic data were generated from six widely used benchmark models of rhythmic and non-rhythmic gene regulatory systems with random perturbations mimicking the effect of gene knock-out or chemical treatments. Then, time-series data of increasing lengths were provided to five state-of-the-art network inference algorithms representing distinctive mathematical paradigms. The performances of such network reconstruction methodologies are uncovered under various experimental conditions. We report that the algorithms do not benefit equally from data increments. Furthermore, for rhythmic systems, it is more profitable for network inference strategies to be run on long time-series rather than short time-series with multiple perturbations. By contrast, for the non-rhythmic systems, increasing the number of perturbation experiments yielded better results than increasing the sampling frequency. We expect that future benchmark and algorithm design would integrate such multifactorial considerations to promote their widespread and conscientious usage.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper shows via simulations on standard benchmarks that long time series beat extra perturbations for rhythmic GRN inference while the reverse holds for non-rhythmic systems, but offers no check that the benchmarks match real data statistics.

read the letter

The central finding is that the five algorithms tested do not improve uniformly with more data, and the preferred regime splits cleanly by system type: longer series help rhythmic models more, while extra perturbations help non-rhythmic ones more. This comes from generating time series from six established benchmark models with random knock-out style perturbations and feeding increasing lengths or numbers of series to the algorithms. The setup is straightforward and directly addresses experimental design tradeoffs that most single-factor benchmarks ignore, which is the useful part. It gives concrete, if simulation-bound, advice on where to spend limited experimental resources. The weakness is exactly the one flagged in the stress test. The performance differences are observed only on these particular models under this perturbation scheme, with no quantitative comparison of their oscillation stability, noise structure, or response distributions to real gene-regulatory time series. Without that match, it is unclear whether the rhythmic/non-rhythmic split will appear in actual experiments. The abstract also gives only directional statements and no error bars or statistical tests, so the size of the reported effects is hard to judge. The work is aimed at computational systems biologists who run or design GRN inference studies and need to think about data collection strategy. It is honest about its simulation basis and engages the literature on benchmarks, so it shows clear thinking even if the generalization step is untested. I would send it to peer review so the full quantitative results and any additional validation can be examined.

Referee Report

2 major / 0 minor

Summary. The paper introduces a multifactorial evaluation framework for assessing five state-of-the-art GRN inference algorithms on simulated time-series data from six benchmark models (rhythmic and non-rhythmic). Realistic data are generated via random perturbations mimicking knock-outs or chemical treatments. Time-series of increasing lengths are supplied to the algorithms, and performance is compared across varying data quantities and perturbation numbers. The central claim is that long time-series outperform short series with multiple perturbations for rhythmic systems, while the reverse holds for non-rhythmic systems; algorithms do not benefit equally from data increments.

Significance. If the results hold, the work supplies pragmatic guidelines for experimental design in GRN reconstruction by quantifying tradeoffs between sampling frequency and perturbation experiments. Strengths include the use of established independent benchmark models, random perturbations that avoid self-referential fitting, and a reproducible simulation-based setup that enables direct comparison of algorithmic paradigms. These elements support falsifiable, multifactorial evaluation. However, the absence of quantitative metrics, error bars, or statistical tests for the reported directional trends, together with the untested mapping from benchmark statistics to real GRN time series, reduces the immediate impact and generalizability.

major comments (2)

[Abstract] Abstract: the directional claims (long time-series preferable for rhythmic systems; more perturbations preferable for non-rhythmic systems) are stated without any quantitative performance metrics (e.g., AUROC, AUPR, or precision-recall values), error bars, or statistical tests across the five algorithms and six models. This omission makes the magnitude, consistency, and reliability of the reported tradeoffs impossible to assess from the provided evidence.
[Results] Results/Discussion: the generalization that the observed rhythmic vs non-rhythmic tradeoff will inform real experimental design rests on the unverified assumption that the six benchmark models plus the specific random perturbation scheme reproduce the relevant dynamical features (oscillation stability, perturbation response distributions, noise structure) of actual gene-regulatory time series. No quantitative comparison of these statistical properties between simulated and experimental data is reported.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major comment below, indicating planned revisions where appropriate.

read point-by-point responses

Referee: [Abstract] Abstract: the directional claims (long time-series preferable for rhythmic systems; more perturbations preferable for non-rhythmic systems) are stated without any quantitative performance metrics (e.g., AUROC, AUPR, or precision-recall values), error bars, or statistical tests across the five algorithms and six models. This omission makes the magnitude, consistency, and reliability of the reported tradeoffs impossible to assess from the provided evidence.

Authors: We agree that the abstract would benefit from quantitative support for the directional claims. In the revised manuscript we will include representative AUROC and AUPR values (with brief indication of variability across algorithms and models) to convey the magnitude and consistency of the observed tradeoffs while preserving conciseness. revision: yes
Referee: [Results] Results/Discussion: the generalization that the observed rhythmic vs non-rhythmic tradeoff will inform real experimental design rests on the unverified assumption that the six benchmark models plus the specific random perturbation scheme reproduce the relevant dynamical features (oscillation stability, perturbation response distributions, noise structure) of actual gene-regulatory time series. No quantitative comparison of these statistical properties between simulated and experimental data is reported.

Authors: The study deliberately employs established, independent benchmark models under controlled perturbations to isolate the effects of data quantity and perturbation number in a reproducible manner. We do not assert that the simulations identically reproduce every statistical feature of real data; rather, they provide a standardized testbed for evaluating algorithmic behavior. We will add an explicit limitations paragraph in the discussion acknowledging that future work should validate the trends on experimental time series. revision: partial

Circularity Check

0 steps flagged

No circularity: performance rankings derived from independent benchmark simulations

full rationale

The paper generates time-series data from six established external benchmark models of GRNs (rhythmic and non-rhythmic), applies random perturbations, and measures inference algorithm performance under varying data regimes. All reported tradeoffs (longer series vs. more perturbations) are direct empirical outcomes on these fixed external models; no parameters are fitted to the inference results themselves, no self-definitional equations appear, and no load-bearing claims reduce to self-citations. The evaluation is self-contained against the chosen benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The evaluation framework rests on the fidelity of six pre-existing benchmark models and the assumption that random perturbations adequately represent real experimental interventions; no new free parameters or invented entities are introduced.

axioms (2)

domain assumption The six widely used benchmark models of rhythmic and non-rhythmic gene regulatory systems generate data whose dynamics are representative of real biological networks.
Data generation step described in abstract.
domain assumption Random perturbations applied to the models mimic the effects of gene knock-out or chemical treatments in real experiments.
Used to create realistic data for the evaluation.

pith-pipeline@v0.9.0 · 5772 in / 1355 out tokens · 30615 ms · 2026-05-25T13:29:48.392217+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

[1]

Aalto, A., Viitasaari, L., Ilmonen, P., and Gon c alves, J. (2018). Continuous time G aussian process dynamical models in gene regulatory network inference. ArXiv:1808.08161

work page arXiv 2018
[2]

Aderhold, A., Husmeier, D., and Grzegorczyk, M. (2017). Approximate B ayesian inference in semi-mechanistic models. Statistics and Computing, 27(4), 1003--1040

work page 2017
[3]

Aderhold, A., Husmeier, D., and Grzegorczyk, M. (2014). Statistical inference of regulatory networks for circadian regulation . Statistical Applications in Genetics and Molecular Biology, 13(3), 227--273. doi:10.1515/sagmb-2013-0051

work page doi:10.1515/sagmb-2013-0051 2014
[4]

Casadiego, J., Nitzan, M., Hallerberg, S., and Timme, M. (2017). Model-free inference of direct network interactions from nonlinear collective dynamics. Nature Communications, 8(1), 2192

work page 2017
[5]

Chaitankar, V., Ghosh, P., Perkins, E.J., Gong, P., and Zhang, C. (2010). Time lagged information theoretic approaches to the reverse engineering of gene regulatory networks. BMC Bioinformatics, 11(6), S19

work page 2010
[6]

Geier, F., Timmer, J., and Fleck, C. (2007). Reconstructing gene-regulatory networks from time series, knock-out data, and prior knowledge. BMC systems biology, 1(1), 11

work page 2007
[7]

Gillespie, D. (2000). The chemical L angevin equation. The Journal of Chemical Physics, 113(1), 297--306

work page 2000
[8]

Guerriero, M., Pokhilko, A., Fern\' a ndez, A., Halliday, K., Millar, A., and Hillston, J. (2011). Stochastic properties of the plant circadian clock. Journal of the Royal Society Interface, 9(69), 744--756

work page 2011
[9]

Haque, S., Ahmad, J.S., Clark, N.M., Williams, C.M., and Sozzani, R. (2019). Computational prediction of gene regulatory networks in plant growth and development. Current Opinion in Plant Biology, 47, 96 -- 105. doi:https://doi.org/10.1016/j.pbi.2018.10.005. Growth and development

work page doi:10.1016/j.pbi.2018.10.005 2019
[10]

and Geurts, P

Huynh-Thu , V. and Geurts, P. (2018). dynGENIE3 : dynamical GENIE3 for the inference of gene networks from time series expression data. Nature Scientific Reports, 8(1), 3384

work page 2018
[11]

and Geurts, P

Huynh-Thu, V.A. and Geurts, P. (2018). dynGENIE3: dynamical GENIE3 for the inference of gene networks from time series expression data . Scientific Reports, 8(1), 3384. doi:10.1038/s41598-018-21715-0

work page doi:10.1038/s41598-018-21715-0 2018
[12]

Madhamshettiwar, P.B., Maetschke, S.R., Davis, M.J., Reverter, A., and Ragan, M.A. (2012). Gene regulatory network inference: evaluation and application to ovarian cancer allows the prioritization of drug targets. Genome Medicine, 4(5), 41

work page 2012
[13]

Marbach, D., Prill, R., Schaffter, T., Mattiussi, C., Floreano, D., and Stolovitzky, G. (2010). Revealing strengths and weaknesses of methods for gene network inference. PNAS, 107(14), 6286--6291

work page 2010
[14]

Marbach, D., Schaffter, T., Mattiussi, C., and Floreano, D. (2009). Generating realistic in silico gene networks for performance assessment of reverse engineering methods. Journal of Computational Biology, 16(2), 229--239

work page 2009
[15]

Marbach, D., Costello, J.C., K \" u ffner, R., Vega, N.M., Prill, R.J., Camacho, D.M., Allison, K.R., Kellis, M., Collins, J.J., Aderhold, A., Stolovitzky, G., Bonneau, R., Chen, Y., Cordero, F., Crane, M., Dondelinger, F., Drton, M., Esposito, R., Foygel, R., De La Fuente , A., Gertheiss, J., Geurts, P., Greenfield, A., Grzegorczyk, M., Haury, A.C., Holm...

work page doi:10.1038/nmeth.2016 2012
[16]

Markdahl, J., Colombo, N., Thunberg, J., and Goncalves, J. (2017). Experimental design tradeoffs for gene regulatory network inference: An in silico study of the yeast S accharomyces cerevisiae cell cycle. IEEE Conference on Decision and Control (CDC), 423--428

work page 2017
[17]

Mombaerts, L., Carignano, A., Robertson, F., Hearn, T., Junyang, J., Hayden, D., Rutterford, Z., Hotta, C., Hubbard, K., Maria, M., Yuan, Y., Hannah, M., Goncalves, J., and Webb, A. (2019). Dynamical differential expression ( DyDE ) reveals the period control mechanisms of the A rabidopsis circadian oscillator. PLoS Computational Biology, 15(1), e1006674

work page 2019
[18]

Mombaerts, L., Mauroy, A., and Goncalves, J. (2016). Optimising time-series experimental design for modelling of circadian rhythms: the value of transient data. IFAC-PapersOnLine, 49(26), 109--113

work page 2016
[19]

Muldoon, J., Yu, J., Fassia, M., and Bagheri, N. (2019). Network inference performance complexity: a consequence of topological, experimental, and algorithmic determinants. Bioinformatics. doi:10.1093/bioinformatics/btz105

work page doi:10.1093/bioinformatics/btz105 2019
[20]

Pokhilko, A., Hodge, S.K., Stratford, K., Knox, K., Edwards, K.D., Thomson, A.W., Mizuno, T., and Millar, A.J. (2010). Data assimilation constrains new connections and components in a complex, eukaryotic circadian clock model. Molecular Systems Biology, 6(1). doi:10.1038/msb.2010.69

work page doi:10.1038/msb.2010.69 2010
[21]

Prill, R., Marbach, D., Saez-Rodriguez, J., Sorger, P., Alexopoulos, L., Xue, X., Clarke, N., Altan-Bonnet, G., and Stolovitzky, G. (2010). Towards a rigorous assessment of systems biology models: the DREAM3 challenges. PLoS ONE, 5(2), e9202

work page 2010
[22]

Schaffter, T., Marbach, D., and Floreano, D. (2011). GeneNetWeaver: In silico benchmark generation and performance profiling of network inference methods . Bioinformatics, 27(16), 2263--2270. doi:10.1093/bioinformatics/btr373

work page doi:10.1093/bioinformatics/btr373 2011
[23]

Sefer, E., Kleyman, M., and Bar-Joseph , Z. (2016). Tradeoffs between dense and replicate sampling strategies for high-throughput time series experiments. Cell Systems, 3, 35--42

work page 2016
[24]

Tibshirani, R. (1996). Regression shrinkage and selection via the L asso. Journal of the Royal Statistical Society, Ser. B, 58, 267--288

work page 1996
[25]

and Gunawan, R

Ud-Dean, S.M. and Gunawan, R. (2015). Optimal design of gene knockout experiments for gene regulatory network inference . Bioinformatics, 32(6), 875--883. doi:10.1093/bioinformatics/btv672

work page doi:10.1093/bioinformatics/btv672 2015

[1] [1]

Aalto, A., Viitasaari, L., Ilmonen, P., and Gon c alves, J. (2018). Continuous time G aussian process dynamical models in gene regulatory network inference. ArXiv:1808.08161

work page arXiv 2018

[2] [2]

Aderhold, A., Husmeier, D., and Grzegorczyk, M. (2017). Approximate B ayesian inference in semi-mechanistic models. Statistics and Computing, 27(4), 1003--1040

work page 2017

[3] [3]

Aderhold, A., Husmeier, D., and Grzegorczyk, M. (2014). Statistical inference of regulatory networks for circadian regulation . Statistical Applications in Genetics and Molecular Biology, 13(3), 227--273. doi:10.1515/sagmb-2013-0051

work page doi:10.1515/sagmb-2013-0051 2014

[4] [4]

Casadiego, J., Nitzan, M., Hallerberg, S., and Timme, M. (2017). Model-free inference of direct network interactions from nonlinear collective dynamics. Nature Communications, 8(1), 2192

work page 2017

[5] [5]

Chaitankar, V., Ghosh, P., Perkins, E.J., Gong, P., and Zhang, C. (2010). Time lagged information theoretic approaches to the reverse engineering of gene regulatory networks. BMC Bioinformatics, 11(6), S19

work page 2010

[6] [6]

Geier, F., Timmer, J., and Fleck, C. (2007). Reconstructing gene-regulatory networks from time series, knock-out data, and prior knowledge. BMC systems biology, 1(1), 11

work page 2007

[7] [7]

Gillespie, D. (2000). The chemical L angevin equation. The Journal of Chemical Physics, 113(1), 297--306

work page 2000

[8] [8]

Guerriero, M., Pokhilko, A., Fern\' a ndez, A., Halliday, K., Millar, A., and Hillston, J. (2011). Stochastic properties of the plant circadian clock. Journal of the Royal Society Interface, 9(69), 744--756

work page 2011

[9] [9]

Haque, S., Ahmad, J.S., Clark, N.M., Williams, C.M., and Sozzani, R. (2019). Computational prediction of gene regulatory networks in plant growth and development. Current Opinion in Plant Biology, 47, 96 -- 105. doi:https://doi.org/10.1016/j.pbi.2018.10.005. Growth and development

work page doi:10.1016/j.pbi.2018.10.005 2019

[10] [10]

and Geurts, P

Huynh-Thu , V. and Geurts, P. (2018). dynGENIE3 : dynamical GENIE3 for the inference of gene networks from time series expression data. Nature Scientific Reports, 8(1), 3384

work page 2018

[11] [11]

and Geurts, P

Huynh-Thu, V.A. and Geurts, P. (2018). dynGENIE3: dynamical GENIE3 for the inference of gene networks from time series expression data . Scientific Reports, 8(1), 3384. doi:10.1038/s41598-018-21715-0

work page doi:10.1038/s41598-018-21715-0 2018

[12] [12]

Madhamshettiwar, P.B., Maetschke, S.R., Davis, M.J., Reverter, A., and Ragan, M.A. (2012). Gene regulatory network inference: evaluation and application to ovarian cancer allows the prioritization of drug targets. Genome Medicine, 4(5), 41

work page 2012

[13] [13]

Marbach, D., Prill, R., Schaffter, T., Mattiussi, C., Floreano, D., and Stolovitzky, G. (2010). Revealing strengths and weaknesses of methods for gene network inference. PNAS, 107(14), 6286--6291

work page 2010

[14] [14]

Marbach, D., Schaffter, T., Mattiussi, C., and Floreano, D. (2009). Generating realistic in silico gene networks for performance assessment of reverse engineering methods. Journal of Computational Biology, 16(2), 229--239

work page 2009

[15] [15]

Marbach, D., Costello, J.C., K \" u ffner, R., Vega, N.M., Prill, R.J., Camacho, D.M., Allison, K.R., Kellis, M., Collins, J.J., Aderhold, A., Stolovitzky, G., Bonneau, R., Chen, Y., Cordero, F., Crane, M., Dondelinger, F., Drton, M., Esposito, R., Foygel, R., De La Fuente , A., Gertheiss, J., Geurts, P., Greenfield, A., Grzegorczyk, M., Haury, A.C., Holm...

work page doi:10.1038/nmeth.2016 2012

[16] [16]

Markdahl, J., Colombo, N., Thunberg, J., and Goncalves, J. (2017). Experimental design tradeoffs for gene regulatory network inference: An in silico study of the yeast S accharomyces cerevisiae cell cycle. IEEE Conference on Decision and Control (CDC), 423--428

work page 2017

[17] [17]

Mombaerts, L., Carignano, A., Robertson, F., Hearn, T., Junyang, J., Hayden, D., Rutterford, Z., Hotta, C., Hubbard, K., Maria, M., Yuan, Y., Hannah, M., Goncalves, J., and Webb, A. (2019). Dynamical differential expression ( DyDE ) reveals the period control mechanisms of the A rabidopsis circadian oscillator. PLoS Computational Biology, 15(1), e1006674

work page 2019

[18] [18]

Mombaerts, L., Mauroy, A., and Goncalves, J. (2016). Optimising time-series experimental design for modelling of circadian rhythms: the value of transient data. IFAC-PapersOnLine, 49(26), 109--113

work page 2016

[19] [19]

Muldoon, J., Yu, J., Fassia, M., and Bagheri, N. (2019). Network inference performance complexity: a consequence of topological, experimental, and algorithmic determinants. Bioinformatics. doi:10.1093/bioinformatics/btz105

work page doi:10.1093/bioinformatics/btz105 2019

[20] [20]

Pokhilko, A., Hodge, S.K., Stratford, K., Knox, K., Edwards, K.D., Thomson, A.W., Mizuno, T., and Millar, A.J. (2010). Data assimilation constrains new connections and components in a complex, eukaryotic circadian clock model. Molecular Systems Biology, 6(1). doi:10.1038/msb.2010.69

work page doi:10.1038/msb.2010.69 2010

[21] [21]

Prill, R., Marbach, D., Saez-Rodriguez, J., Sorger, P., Alexopoulos, L., Xue, X., Clarke, N., Altan-Bonnet, G., and Stolovitzky, G. (2010). Towards a rigorous assessment of systems biology models: the DREAM3 challenges. PLoS ONE, 5(2), e9202

work page 2010

[22] [22]

Schaffter, T., Marbach, D., and Floreano, D. (2011). GeneNetWeaver: In silico benchmark generation and performance profiling of network inference methods . Bioinformatics, 27(16), 2263--2270. doi:10.1093/bioinformatics/btr373

work page doi:10.1093/bioinformatics/btr373 2011

[23] [23]

Sefer, E., Kleyman, M., and Bar-Joseph , Z. (2016). Tradeoffs between dense and replicate sampling strategies for high-throughput time series experiments. Cell Systems, 3, 35--42

work page 2016

[24] [24]

Tibshirani, R. (1996). Regression shrinkage and selection via the L asso. Journal of the Royal Statistical Society, Ser. B, 58, 267--288

work page 1996

[25] [25]

and Gunawan, R

Ud-Dean, S.M. and Gunawan, R. (2015). Optimal design of gene knockout experiments for gene regulatory network inference . Bioinformatics, 32(6), 875--883. doi:10.1093/bioinformatics/btv672

work page doi:10.1093/bioinformatics/btv672 2015