Make the most of your samples: Bayes factor estimators for high-dimensional models of sequence evolution View Full Text


Ontology type: schema:ScholarlyArticle      Open Access: True


Article Info

DATE

2013-12

AUTHORS

Guy Baele, Philippe Lemey, Stijn Vansteelandt

ABSTRACT

BACKGROUND: Accurate model comparison requires extensive computation times, especially for parameter-rich models of sequence evolution. In the Bayesian framework, model selection is typically performed through the evaluation of a Bayes factor, the ratio of two marginal likelihoods (one for each model). Recently introduced techniques to estimate (log) marginal likelihoods, such as path sampling and stepping-stone sampling, offer increased accuracy over the traditional harmonic mean estimator at an increased computational cost. Most often, each model's marginal likelihood will be estimated individually, which leads the resulting Bayes factor to suffer from errors associated with each of these independent estimation processes. RESULTS: We here assess the original 'model-switch' path sampling approach for direct Bayes factor estimation in phylogenetics, as well as an extension that uses more samples, to construct a direct path between two competing models, thereby eliminating the need to calculate each model's marginal likelihood independently. Further, we provide a competing Bayes factor estimator using an adaptation of the recently introduced stepping-stone sampling algorithm and set out to determine appropriate settings for accurately calculating such Bayes factors, with context-dependent evolutionary models as an example. While we show that modest efforts are required to roughly identify the increase in model fit, only drastically increased computation times ensure the accuracy needed to detect more subtle details of the evolutionary process. CONCLUSIONS: We show that our adaptation of stepping-stone sampling for direct Bayes factor calculation outperforms the original path sampling approach as well as an extension that exploits more samples. Our proposed approach for Bayes factor estimation also has preferable statistical properties over the use of individual marginal likelihood estimates for both models under comparison. Assuming a sigmoid function to determine the path between two competing models, we provide evidence that a single well-chosen sigmoid shape value requires less computational efforts in order to approximate the true value of the (log) Bayes factor compared to the original approach. We show that the (log) Bayes factors calculated using path sampling and stepping-stone sampling differ drastically from those estimated using either of the harmonic mean estimators, supporting earlier claims that the latter systematically overestimate the performance of high-dimensional models, which we show can lead to erroneous conclusions. Based on our results, we argue that highly accurate estimation of differences in model fit for high-dimensional models requires much more computational effort than suggested in recent studies on marginal likelihood estimation. More... »

PAGES

85

Identifiers

URI

http://scigraph.springernature.com/pub.10.1186/1471-2105-14-85

DOI

http://dx.doi.org/10.1186/1471-2105-14-85

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1043342329

PUBMED

https://www.ncbi.nlm.nih.gov/pubmed/23497171


Indexing Status Check whether this publication has been indexed by Scopus and Web Of Science using the SN Indexing Status Tool
Incoming Citations Browse incoming citations for this publication using opencitations.net

JSON-LD is the canonical representation for SciGraph data.

TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

[
  {
    "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
    "about": [
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/0104", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Statistics", 
        "type": "DefinedTerm"
      }, 
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/01", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Mathematical Sciences", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Algorithms", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Animals", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Bayes Theorem", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Evolution, Molecular", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Genes, rRNA", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Humans", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Likelihood Functions", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Models, Genetic", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Phylogeny", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Pseudogenes", 
        "type": "DefinedTerm"
      }, 
      {
        "inDefinedTermSet": "https://www.nlm.nih.gov/mesh/", 
        "name": "Sequence Analysis, DNA", 
        "type": "DefinedTerm"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "alternateName": "KU Leuven", 
          "id": "https://www.grid.ac/institutes/grid.5596.f", 
          "name": [
            "Department of Microbiology and Immunology, Rega Institute, KU Leuven, Kapucijnen-voer 33 blok I bus 7001, B-3000, Leuven, Belgium"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Baele", 
        "givenName": "Guy", 
        "id": "sg:person.01211561143.07", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01211561143.07"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "KU Leuven", 
          "id": "https://www.grid.ac/institutes/grid.5596.f", 
          "name": [
            "Department of Microbiology and Immunology, Rega Institute, KU Leuven, Kapucijnen-voer 33 blok I bus 7001, B-3000, Leuven, Belgium"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Lemey", 
        "givenName": "Philippe", 
        "id": "sg:person.01261447574.28", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01261447574.28"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Ghent University", 
          "id": "https://www.grid.ac/institutes/grid.5342.0", 
          "name": [
            "Department of Applied Mathematics and Computer Science, Ghent University, Krijgslaan 281 S9, B-9000, Ghent, Belgium"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Vansteelandt", 
        "givenName": "Stijn", 
        "id": "sg:person.0637036363.16", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0637036363.16"
        ], 
        "type": "Person"
      }
    ], 
    "citation": [
      {
        "id": "https://doi.org/10.1093/oxfordjournals.molbev.a003872", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1004809702"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1186/1471-2148-10-244", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1005139630", 
          "https://doi.org/10.1186/1471-2148-10-244"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1186/1471-2148-9-87", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1010845412", 
          "https://doi.org/10.1186/1471-2148-9-87"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/bf00178256", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1012226005", 
          "https://doi.org/10.1007/bf00178256"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1016/j.tig.2005.04.001", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1013335821"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/mss075", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1013930907"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1016/0169-5347(96)10041-0", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1014862677"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/oxfordjournals.molbev.a025811", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1016347110"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/sysbio/syq085", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1020904512"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s11692-011-9139-2", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1021886860", 
          "https://doi.org/10.1007/s11692-011-9139-2"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/msh112", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1022502093"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/bf01406511", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1024531907", 
          "https://doi.org/10.1007/bf01406511"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/bf01406511", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1024531907", 
          "https://doi.org/10.1007/bf01406511"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/mss243", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1029511886"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/msl041", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1034128209"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/mss084", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1041314909"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/msq224", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1041482208"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1093/molbev/msn104", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1044907967"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/10635150490264699", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1046054260"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00239-010-9362-y", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1047245627", 
          "https://doi.org/10.1007/s00239-010-9362-y"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00239-010-9362-y", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1047245627", 
          "https://doi.org/10.1007/s00239-010-9362-y"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1073/pnas.0404142101", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1048935256"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1017/s030500410001330x", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1053861300"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/01621459.1995.10476572", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058304855"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/01621459.1996.10476995", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058305129"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/01621459.1997.10474045", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058305294"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/106351502753475862", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058369254"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/10635150490522584", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058369478"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/10635150500433722", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058369524"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/10635150590945313", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058369574"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/10635150590947041", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058369580"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1080/10635150802422324", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1058369811"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1126/science.1065156", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1062445470"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1214/ss/1028905934", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1064409457"
        ], 
        "type": "CreativeWork"
      }
    ], 
    "datePublished": "2013-12", 
    "datePublishedReg": "2013-12-01", 
    "description": "BACKGROUND: Accurate model comparison requires extensive computation times, especially for parameter-rich models of sequence evolution. In the Bayesian framework, model selection is typically performed through the evaluation of a Bayes factor, the ratio of two marginal likelihoods (one for each model). Recently introduced techniques to estimate (log) marginal likelihoods, such as path sampling and stepping-stone sampling, offer increased accuracy over the traditional harmonic mean estimator at an increased computational cost. Most often, each model's marginal likelihood will be estimated individually, which leads the resulting Bayes factor to suffer from errors associated with each of these independent estimation processes.\nRESULTS: We here assess the original 'model-switch' path sampling approach for direct Bayes factor estimation in phylogenetics, as well as an extension that uses more samples, to construct a direct path between two competing models, thereby eliminating the need to calculate each model's marginal likelihood independently. Further, we provide a competing Bayes factor estimator using an adaptation of the recently introduced stepping-stone sampling algorithm and set out to determine appropriate settings for accurately calculating such Bayes factors, with context-dependent evolutionary models as an example. While we show that modest efforts are required to roughly identify the increase in model fit, only drastically increased computation times ensure the accuracy needed to detect more subtle details of the evolutionary process.\nCONCLUSIONS: We show that our adaptation of stepping-stone sampling for direct Bayes factor calculation outperforms the original path sampling approach as well as an extension that exploits more samples. Our proposed approach for Bayes factor estimation also has preferable statistical properties over the use of individual marginal likelihood estimates for both models under comparison. Assuming a sigmoid function to determine the path between two competing models, we provide evidence that a single well-chosen sigmoid shape value requires less computational efforts in order to approximate the true value of the (log) Bayes factor compared to the original approach. We show that the (log) Bayes factors calculated using path sampling and stepping-stone sampling differ drastically from those estimated using either of the harmonic mean estimators, supporting earlier claims that the latter systematically overestimate the performance of high-dimensional models, which we show can lead to erroneous conclusions. Based on our results, we argue that highly accurate estimation of differences in model fit for high-dimensional models requires much more computational effort than suggested in recent studies on marginal likelihood estimation.", 
    "genre": "research_article", 
    "id": "sg:pub.10.1186/1471-2105-14-85", 
    "inLanguage": [
      "en"
    ], 
    "isAccessibleForFree": true, 
    "isFundedItemOf": [
      {
        "id": "sg:grant.3786570", 
        "type": "MonetaryGrant"
      }, 
      {
        "id": "sg:grant.3786660", 
        "type": "MonetaryGrant"
      }
    ], 
    "isPartOf": [
      {
        "id": "sg:journal.1023786", 
        "issn": [
          "1471-2105"
        ], 
        "name": "BMC Bioinformatics", 
        "type": "Periodical"
      }, 
      {
        "issueNumber": "1", 
        "type": "PublicationIssue"
      }, 
      {
        "type": "PublicationVolume", 
        "volumeNumber": "14"
      }
    ], 
    "name": "Make the most of your samples: Bayes factor estimators for high-dimensional models of sequence evolution", 
    "pagination": "85", 
    "productId": [
      {
        "name": "readcube_id", 
        "type": "PropertyValue", 
        "value": [
          "ce74e8ce69ad33170375275efb18b8b9f7f7e9c14c77ff8f0e1be985e2d64325"
        ]
      }, 
      {
        "name": "pubmed_id", 
        "type": "PropertyValue", 
        "value": [
          "23497171"
        ]
      }, 
      {
        "name": "nlm_unique_id", 
        "type": "PropertyValue", 
        "value": [
          "100965194"
        ]
      }, 
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1186/1471-2105-14-85"
        ]
      }, 
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1043342329"
        ]
      }
    ], 
    "sameAs": [
      "https://doi.org/10.1186/1471-2105-14-85", 
      "https://app.dimensions.ai/details/publication/pub.1043342329"
    ], 
    "sdDataset": "articles", 
    "sdDatePublished": "2019-04-10T17:30", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-uberresearch-data-dimensions-target-20181106-alternative/cleanup/v134/2549eaecd7973599484d7c17b260dba0a4ecb94b/merge/v9/a6c9fde33151104705d4d7ff012ea9563521a3ce/jats-lookup/v90/0000000001_0000000264/records_8672_00000507.jsonl", 
    "type": "ScholarlyArticle", 
    "url": "http://link.springer.com/10.1186%2F1471-2105-14-85"
  }
]
 

Download the RDF metadata as:  json-ld nt turtle xml License info

HOW TO GET THIS DATA PROGRAMMATICALLY:

JSON-LD is a popular format for linked data which is fully compatible with JSON.

curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/pub.10.1186/1471-2105-14-85'

N-Triples is a line-based linked data format ideal for batch operations.

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1186/1471-2105-14-85'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1186/1471-2105-14-85'

RDF/XML is a standard XML format for linked data.

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1186/1471-2105-14-85'


 

This table displays all metadata directly associated to this object as RDF triples.

235 TRIPLES      21 PREDICATES      72 URIs      32 LITERALS      20 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1186/1471-2105-14-85 schema:about N02b9b5d9612f443ba2172a82d340993e
2 N1505ae2d5e044efe823ad6b236bf7231
3 N9522dd63ff5841238f49aa66a80ff6d8
4 Na607dc00a2bb4c34bf15b4e7bbf20876
5 Nb20b07d70b3e48e0b219ed73ec32f1ec
6 Nc0443d774624434f88eac339e3be7b34
7 Nc49cb8d4ce184fcb8f90743c3d43965a
8 Ncddab52afc3e4da78be696935b0f3911
9 Ne0b3caaee8774d7e8f6f9529797a9597
10 Ne8184da62afd4ff4b86545f0da7479ae
11 Nfce43a5d382c43498cf9cebf55fe89d1
12 anzsrc-for:01
13 anzsrc-for:0104
14 schema:author Nd164860bccdf4368abfa07a1c686815a
15 schema:citation sg:pub.10.1007/bf00178256
16 sg:pub.10.1007/bf01406511
17 sg:pub.10.1007/s00239-010-9362-y
18 sg:pub.10.1007/s11692-011-9139-2
19 sg:pub.10.1186/1471-2148-10-244
20 sg:pub.10.1186/1471-2148-9-87
21 https://doi.org/10.1016/0169-5347(96)10041-0
22 https://doi.org/10.1016/j.tig.2005.04.001
23 https://doi.org/10.1017/s030500410001330x
24 https://doi.org/10.1073/pnas.0404142101
25 https://doi.org/10.1080/01621459.1995.10476572
26 https://doi.org/10.1080/01621459.1996.10476995
27 https://doi.org/10.1080/01621459.1997.10474045
28 https://doi.org/10.1080/106351502753475862
29 https://doi.org/10.1080/10635150490264699
30 https://doi.org/10.1080/10635150490522584
31 https://doi.org/10.1080/10635150500433722
32 https://doi.org/10.1080/10635150590945313
33 https://doi.org/10.1080/10635150590947041
34 https://doi.org/10.1080/10635150802422324
35 https://doi.org/10.1093/molbev/msh112
36 https://doi.org/10.1093/molbev/msl041
37 https://doi.org/10.1093/molbev/msn104
38 https://doi.org/10.1093/molbev/msq224
39 https://doi.org/10.1093/molbev/mss075
40 https://doi.org/10.1093/molbev/mss084
41 https://doi.org/10.1093/molbev/mss243
42 https://doi.org/10.1093/oxfordjournals.molbev.a003872
43 https://doi.org/10.1093/oxfordjournals.molbev.a025811
44 https://doi.org/10.1093/sysbio/syq085
45 https://doi.org/10.1126/science.1065156
46 https://doi.org/10.1214/ss/1028905934
47 schema:datePublished 2013-12
48 schema:datePublishedReg 2013-12-01
49 schema:description BACKGROUND: Accurate model comparison requires extensive computation times, especially for parameter-rich models of sequence evolution. In the Bayesian framework, model selection is typically performed through the evaluation of a Bayes factor, the ratio of two marginal likelihoods (one for each model). Recently introduced techniques to estimate (log) marginal likelihoods, such as path sampling and stepping-stone sampling, offer increased accuracy over the traditional harmonic mean estimator at an increased computational cost. Most often, each model's marginal likelihood will be estimated individually, which leads the resulting Bayes factor to suffer from errors associated with each of these independent estimation processes. RESULTS: We here assess the original 'model-switch' path sampling approach for direct Bayes factor estimation in phylogenetics, as well as an extension that uses more samples, to construct a direct path between two competing models, thereby eliminating the need to calculate each model's marginal likelihood independently. Further, we provide a competing Bayes factor estimator using an adaptation of the recently introduced stepping-stone sampling algorithm and set out to determine appropriate settings for accurately calculating such Bayes factors, with context-dependent evolutionary models as an example. While we show that modest efforts are required to roughly identify the increase in model fit, only drastically increased computation times ensure the accuracy needed to detect more subtle details of the evolutionary process. CONCLUSIONS: We show that our adaptation of stepping-stone sampling for direct Bayes factor calculation outperforms the original path sampling approach as well as an extension that exploits more samples. Our proposed approach for Bayes factor estimation also has preferable statistical properties over the use of individual marginal likelihood estimates for both models under comparison. Assuming a sigmoid function to determine the path between two competing models, we provide evidence that a single well-chosen sigmoid shape value requires less computational efforts in order to approximate the true value of the (log) Bayes factor compared to the original approach. We show that the (log) Bayes factors calculated using path sampling and stepping-stone sampling differ drastically from those estimated using either of the harmonic mean estimators, supporting earlier claims that the latter systematically overestimate the performance of high-dimensional models, which we show can lead to erroneous conclusions. Based on our results, we argue that highly accurate estimation of differences in model fit for high-dimensional models requires much more computational effort than suggested in recent studies on marginal likelihood estimation.
50 schema:genre research_article
51 schema:inLanguage en
52 schema:isAccessibleForFree true
53 schema:isPartOf N05aa3e803cbb405daa274f4e7acbc4df
54 Necf0d7ce7a594de49f643edb9c05c041
55 sg:journal.1023786
56 schema:name Make the most of your samples: Bayes factor estimators for high-dimensional models of sequence evolution
57 schema:pagination 85
58 schema:productId N3b2118df930e4719b1775af04e4aec31
59 N4f3b66cfd28e408b91ec1e370802a1d8
60 Nbef47aa7b3a248d9acfe56788e74f8df
61 Nd5b1e1c25ccc4636a20cfdf6e1175b00
62 Nf8ee00d547f84636b84fdd5676dd7ae6
63 schema:sameAs https://app.dimensions.ai/details/publication/pub.1043342329
64 https://doi.org/10.1186/1471-2105-14-85
65 schema:sdDatePublished 2019-04-10T17:30
66 schema:sdLicense https://scigraph.springernature.com/explorer/license/
67 schema:sdPublisher N6350d792cf2044deb2179450857dedeb
68 schema:url http://link.springer.com/10.1186%2F1471-2105-14-85
69 sgo:license sg:explorer/license/
70 sgo:sdDataset articles
71 rdf:type schema:ScholarlyArticle
72 N02b9b5d9612f443ba2172a82d340993e schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
73 schema:name Genes, rRNA
74 rdf:type schema:DefinedTerm
75 N05aa3e803cbb405daa274f4e7acbc4df schema:volumeNumber 14
76 rdf:type schema:PublicationVolume
77 N1505ae2d5e044efe823ad6b236bf7231 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
78 schema:name Models, Genetic
79 rdf:type schema:DefinedTerm
80 N3b2118df930e4719b1775af04e4aec31 schema:name pubmed_id
81 schema:value 23497171
82 rdf:type schema:PropertyValue
83 N448347dc8f874dcc99ace81efd7b7fad rdf:first sg:person.0637036363.16
84 rdf:rest rdf:nil
85 N4b128c0f07b34094bb2500e1d31b2e29 rdf:first sg:person.01261447574.28
86 rdf:rest N448347dc8f874dcc99ace81efd7b7fad
87 N4f3b66cfd28e408b91ec1e370802a1d8 schema:name dimensions_id
88 schema:value pub.1043342329
89 rdf:type schema:PropertyValue
90 N6350d792cf2044deb2179450857dedeb schema:name Springer Nature - SN SciGraph project
91 rdf:type schema:Organization
92 N9522dd63ff5841238f49aa66a80ff6d8 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
93 schema:name Pseudogenes
94 rdf:type schema:DefinedTerm
95 Na607dc00a2bb4c34bf15b4e7bbf20876 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
96 schema:name Humans
97 rdf:type schema:DefinedTerm
98 Nb20b07d70b3e48e0b219ed73ec32f1ec schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
99 schema:name Likelihood Functions
100 rdf:type schema:DefinedTerm
101 Nbef47aa7b3a248d9acfe56788e74f8df schema:name nlm_unique_id
102 schema:value 100965194
103 rdf:type schema:PropertyValue
104 Nc0443d774624434f88eac339e3be7b34 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
105 schema:name Bayes Theorem
106 rdf:type schema:DefinedTerm
107 Nc49cb8d4ce184fcb8f90743c3d43965a schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
108 schema:name Evolution, Molecular
109 rdf:type schema:DefinedTerm
110 Ncddab52afc3e4da78be696935b0f3911 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
111 schema:name Algorithms
112 rdf:type schema:DefinedTerm
113 Nd164860bccdf4368abfa07a1c686815a rdf:first sg:person.01211561143.07
114 rdf:rest N4b128c0f07b34094bb2500e1d31b2e29
115 Nd5b1e1c25ccc4636a20cfdf6e1175b00 schema:name readcube_id
116 schema:value ce74e8ce69ad33170375275efb18b8b9f7f7e9c14c77ff8f0e1be985e2d64325
117 rdf:type schema:PropertyValue
118 Ne0b3caaee8774d7e8f6f9529797a9597 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
119 schema:name Sequence Analysis, DNA
120 rdf:type schema:DefinedTerm
121 Ne8184da62afd4ff4b86545f0da7479ae schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
122 schema:name Animals
123 rdf:type schema:DefinedTerm
124 Necf0d7ce7a594de49f643edb9c05c041 schema:issueNumber 1
125 rdf:type schema:PublicationIssue
126 Nf8ee00d547f84636b84fdd5676dd7ae6 schema:name doi
127 schema:value 10.1186/1471-2105-14-85
128 rdf:type schema:PropertyValue
129 Nfce43a5d382c43498cf9cebf55fe89d1 schema:inDefinedTermSet https://www.nlm.nih.gov/mesh/
130 schema:name Phylogeny
131 rdf:type schema:DefinedTerm
132 anzsrc-for:01 schema:inDefinedTermSet anzsrc-for:
133 schema:name Mathematical Sciences
134 rdf:type schema:DefinedTerm
135 anzsrc-for:0104 schema:inDefinedTermSet anzsrc-for:
136 schema:name Statistics
137 rdf:type schema:DefinedTerm
138 sg:grant.3786570 http://pending.schema.org/fundedItem sg:pub.10.1186/1471-2105-14-85
139 rdf:type schema:MonetaryGrant
140 sg:grant.3786660 http://pending.schema.org/fundedItem sg:pub.10.1186/1471-2105-14-85
141 rdf:type schema:MonetaryGrant
142 sg:journal.1023786 schema:issn 1471-2105
143 schema:name BMC Bioinformatics
144 rdf:type schema:Periodical
145 sg:person.01211561143.07 schema:affiliation https://www.grid.ac/institutes/grid.5596.f
146 schema:familyName Baele
147 schema:givenName Guy
148 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01211561143.07
149 rdf:type schema:Person
150 sg:person.01261447574.28 schema:affiliation https://www.grid.ac/institutes/grid.5596.f
151 schema:familyName Lemey
152 schema:givenName Philippe
153 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01261447574.28
154 rdf:type schema:Person
155 sg:person.0637036363.16 schema:affiliation https://www.grid.ac/institutes/grid.5342.0
156 schema:familyName Vansteelandt
157 schema:givenName Stijn
158 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0637036363.16
159 rdf:type schema:Person
160 sg:pub.10.1007/bf00178256 schema:sameAs https://app.dimensions.ai/details/publication/pub.1012226005
161 https://doi.org/10.1007/bf00178256
162 rdf:type schema:CreativeWork
163 sg:pub.10.1007/bf01406511 schema:sameAs https://app.dimensions.ai/details/publication/pub.1024531907
164 https://doi.org/10.1007/bf01406511
165 rdf:type schema:CreativeWork
166 sg:pub.10.1007/s00239-010-9362-y schema:sameAs https://app.dimensions.ai/details/publication/pub.1047245627
167 https://doi.org/10.1007/s00239-010-9362-y
168 rdf:type schema:CreativeWork
169 sg:pub.10.1007/s11692-011-9139-2 schema:sameAs https://app.dimensions.ai/details/publication/pub.1021886860
170 https://doi.org/10.1007/s11692-011-9139-2
171 rdf:type schema:CreativeWork
172 sg:pub.10.1186/1471-2148-10-244 schema:sameAs https://app.dimensions.ai/details/publication/pub.1005139630
173 https://doi.org/10.1186/1471-2148-10-244
174 rdf:type schema:CreativeWork
175 sg:pub.10.1186/1471-2148-9-87 schema:sameAs https://app.dimensions.ai/details/publication/pub.1010845412
176 https://doi.org/10.1186/1471-2148-9-87
177 rdf:type schema:CreativeWork
178 https://doi.org/10.1016/0169-5347(96)10041-0 schema:sameAs https://app.dimensions.ai/details/publication/pub.1014862677
179 rdf:type schema:CreativeWork
180 https://doi.org/10.1016/j.tig.2005.04.001 schema:sameAs https://app.dimensions.ai/details/publication/pub.1013335821
181 rdf:type schema:CreativeWork
182 https://doi.org/10.1017/s030500410001330x schema:sameAs https://app.dimensions.ai/details/publication/pub.1053861300
183 rdf:type schema:CreativeWork
184 https://doi.org/10.1073/pnas.0404142101 schema:sameAs https://app.dimensions.ai/details/publication/pub.1048935256
185 rdf:type schema:CreativeWork
186 https://doi.org/10.1080/01621459.1995.10476572 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058304855
187 rdf:type schema:CreativeWork
188 https://doi.org/10.1080/01621459.1996.10476995 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058305129
189 rdf:type schema:CreativeWork
190 https://doi.org/10.1080/01621459.1997.10474045 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058305294
191 rdf:type schema:CreativeWork
192 https://doi.org/10.1080/106351502753475862 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058369254
193 rdf:type schema:CreativeWork
194 https://doi.org/10.1080/10635150490264699 schema:sameAs https://app.dimensions.ai/details/publication/pub.1046054260
195 rdf:type schema:CreativeWork
196 https://doi.org/10.1080/10635150490522584 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058369478
197 rdf:type schema:CreativeWork
198 https://doi.org/10.1080/10635150500433722 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058369524
199 rdf:type schema:CreativeWork
200 https://doi.org/10.1080/10635150590945313 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058369574
201 rdf:type schema:CreativeWork
202 https://doi.org/10.1080/10635150590947041 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058369580
203 rdf:type schema:CreativeWork
204 https://doi.org/10.1080/10635150802422324 schema:sameAs https://app.dimensions.ai/details/publication/pub.1058369811
205 rdf:type schema:CreativeWork
206 https://doi.org/10.1093/molbev/msh112 schema:sameAs https://app.dimensions.ai/details/publication/pub.1022502093
207 rdf:type schema:CreativeWork
208 https://doi.org/10.1093/molbev/msl041 schema:sameAs https://app.dimensions.ai/details/publication/pub.1034128209
209 rdf:type schema:CreativeWork
210 https://doi.org/10.1093/molbev/msn104 schema:sameAs https://app.dimensions.ai/details/publication/pub.1044907967
211 rdf:type schema:CreativeWork
212 https://doi.org/10.1093/molbev/msq224 schema:sameAs https://app.dimensions.ai/details/publication/pub.1041482208
213 rdf:type schema:CreativeWork
214 https://doi.org/10.1093/molbev/mss075 schema:sameAs https://app.dimensions.ai/details/publication/pub.1013930907
215 rdf:type schema:CreativeWork
216 https://doi.org/10.1093/molbev/mss084 schema:sameAs https://app.dimensions.ai/details/publication/pub.1041314909
217 rdf:type schema:CreativeWork
218 https://doi.org/10.1093/molbev/mss243 schema:sameAs https://app.dimensions.ai/details/publication/pub.1029511886
219 rdf:type schema:CreativeWork
220 https://doi.org/10.1093/oxfordjournals.molbev.a003872 schema:sameAs https://app.dimensions.ai/details/publication/pub.1004809702
221 rdf:type schema:CreativeWork
222 https://doi.org/10.1093/oxfordjournals.molbev.a025811 schema:sameAs https://app.dimensions.ai/details/publication/pub.1016347110
223 rdf:type schema:CreativeWork
224 https://doi.org/10.1093/sysbio/syq085 schema:sameAs https://app.dimensions.ai/details/publication/pub.1020904512
225 rdf:type schema:CreativeWork
226 https://doi.org/10.1126/science.1065156 schema:sameAs https://app.dimensions.ai/details/publication/pub.1062445470
227 rdf:type schema:CreativeWork
228 https://doi.org/10.1214/ss/1028905934 schema:sameAs https://app.dimensions.ai/details/publication/pub.1064409457
229 rdf:type schema:CreativeWork
230 https://www.grid.ac/institutes/grid.5342.0 schema:alternateName Ghent University
231 schema:name Department of Applied Mathematics and Computer Science, Ghent University, Krijgslaan 281 S9, B-9000, Ghent, Belgium
232 rdf:type schema:Organization
233 https://www.grid.ac/institutes/grid.5596.f schema:alternateName KU Leuven
234 schema:name Department of Microbiology and Immunology, Rega Institute, KU Leuven, Kapucijnen-voer 33 blok I bus 7001, B-3000, Leuven, Belgium
235 rdf:type schema:Organization
 




Preview window. Press ESC to close (or click here)


...