Advanced

Turning vice into virtue : Using Batch-Effects to Detect Errors in Large Genomic Datasets

Mafessoni, Fabrizio; Prasad, Rashmi B LU ; Groop, Leif LU ; Hansson, Ola LU and Prüfer, Kay (2018) In Genome Biology and Evolution 10(10). p.2697-2708
Abstract

It is often unavoidable to combine data from different sequencing centers or sequencing platforms when compiling datasets with a large number of individuals. However, the different data are likely to contain specific systematic errors that will appear as SNPs. Here, we devise a method to detect systematic errors in combined datasetIs. To measure quality differences between individual genomes, we study pairs of variants that reside on different chromosomes and co-occur in individuals. The abundance of these pairs of variants in different genomes is then used to detect systematic errors due to batch effects. Applying our method to the 1000 Genomes dataset, we find that coding regions are enriched for errors, where about 1% of the... (More)

It is often unavoidable to combine data from different sequencing centers or sequencing platforms when compiling datasets with a large number of individuals. However, the different data are likely to contain specific systematic errors that will appear as SNPs. Here, we devise a method to detect systematic errors in combined datasetIs. To measure quality differences between individual genomes, we study pairs of variants that reside on different chromosomes and co-occur in individuals. The abundance of these pairs of variants in different genomes is then used to detect systematic errors due to batch effects. Applying our method to the 1000 Genomes dataset, we find that coding regions are enriched for errors, where about 1% of the higher-frequency variants are predicted to be erroneous, whereas errors outside of coding regions are much rarer (<0.001%).As expected, predicted errors are found less often than other variants in a dataset that was generated with a different sequencing technology, indicating that many of the candidates are indeed errors. However, predicted 1000 Genomes errors are also found in other large datasets; our observation is thus not specific to the 1000 Genomes dataset. Our results show that batch effects can be turned into a virtue by using the resulting variation in large scale datasets to detect systematic errors.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
in
Genome Biology and Evolution
volume
10
issue
10
pages
2697 - 2708
publisher
Oxford University Press
external identifiers
  • scopus:85054893761
ISSN
1759-6653
DOI
10.1093/gbe/evy199
language
English
LU publication?
yes
id
3aeb382c-fa6d-4304-b442-679808487ac4
date added to LUP
2018-09-14 17:55:41
date last changed
2019-01-06 14:05:09
@article{3aeb382c-fa6d-4304-b442-679808487ac4,
  abstract     = {<p>It is often unavoidable to combine data from different sequencing centers or sequencing platforms when compiling datasets with a large number of individuals. However, the different data are likely to contain specific systematic errors that will appear as SNPs. Here, we devise a method to detect systematic errors in combined datasetIs. To measure quality differences between individual genomes, we study pairs of variants that reside on different chromosomes and co-occur in individuals. The abundance of these pairs of variants in different genomes is then used to detect systematic errors due to batch effects. Applying our method to the 1000 Genomes dataset, we find that coding regions are enriched for errors, where about 1% of the higher-frequency variants are predicted to be erroneous, whereas errors outside of coding regions are much rarer (&lt;0.001%).As expected, predicted errors are found less often than other variants in a dataset that was generated with a different sequencing technology, indicating that many of the candidates are indeed errors. However, predicted 1000 Genomes errors are also found in other large datasets; our observation is thus not specific to the 1000 Genomes dataset. Our results show that batch effects can be turned into a virtue by using the resulting variation in large scale datasets to detect systematic errors.</p>},
  author       = {Mafessoni, Fabrizio and Prasad, Rashmi B and Groop, Leif and Hansson, Ola and Prüfer, Kay},
  issn         = {1759-6653},
  language     = {eng},
  month        = {09},
  number       = {10},
  pages        = {2697--2708},
  publisher    = {Oxford University Press},
  series       = {Genome Biology and Evolution},
  title        = {Turning vice into virtue : Using Batch-Effects to Detect Errors in Large Genomic Datasets},
  url          = {http://dx.doi.org/10.1093/gbe/evy199},
  volume       = {10},
  year         = {2018},
}