How Important Is It To Gather New Controls For Whole Genome And Exome Variant Studies?

Question

[Jeremy Leipzig]

I would think that people are more likely to try to use 1000 Genomes and dbSNP frequencies rather than pay for new controls in such experiments.

What are the current best practices and standards for these studies?

Answer 1

[Raony Guimarães]

I would say that you need to get as much controls as possible always ! :)

Every time we have a new exome in our lab we try to use it as a control to analyse the next ones... This really helps to reduce the number of candidate variants specially because our population is not well covered in databases such as 1000Genomes, dbSNP or ESP6500.

Another important thing is to use parents and relatives as controls whenever it's possible, this is another thing that also helps to reduce a lot the number of candidates variants, assuming you are doing exome sequencing for identifying the variant responsible for the disease.

Of course with the time, you can expect that the number of variants eliminated when using your own controls will decrease as the number of variants from your population will be well covered in this databases, but using the parents will always reduce a lot the number of candidate variants! This really depends of the population you are working with and the number of individuals from this population that are present in such databases.

I'm curious to see the difference in reduction of variants in our own individuals when the 1000Genomes project publish another 1000 individuals sequenced by the end of this year.

Answer 2

[DG]

If you read through most of the publications out there when using exomes and genomes for identifying mendelian disease genes you don't strictly need to gather controls the same way you do when looking only at specific mutations or genes. Use of 1000 genomes and the ESP6500 dataset is usually "good enough", assuming you are looking at the right population. If you are dealing with an unrepresented population you may need to gather a local control population.

That said, as you do more and more exomes/genomes you should be building a local database of population controls to check your data against. It helps immensely in screening out the genes or variants that may look promising but are red herrings.