Selection of source sample for DNA extraction
UK Biobank participants provide a wide range of biological samples [4, 5] that are aliquotted into 850 μL, 2D bar-coded micro tubes prior to storage. The buffy coat aliquot, derived from 10 ml of whole blood collected into an ethylenediaminetetraacetic acid (EDTA) vacutainer, was selected for DNA extraction as it was available for the majority of participants and was expected to yield the required 10 ng/μL concentration required for genotyping. Saliva was considered but due to the lower number of participants providing this sample type to the UK Biobank study, this was not used.
During retrieval of buffy coat samples for DNA extraction it was important to maximise the picking speed whilst avoiding clustering of participants by time or date of collection, collection centre, geography (UK Biobank recruited from 22 assessment centres across the UK), or any participant phenotype (typically only 2 assessment centres were represented on each stored plate). An algorithm was developed to pick the samples in a way that increased the number of assessment centres per plate to reduce potential systematic bias from the way the samples were originally collected and processed that could affect downstream genotyping.
DNA extraction
A custom DNA extraction system was developed by the Tecan Integration Group (TIG) [12] to enable DNA extraction and quantification in a single process (Fig. 1). DNA was extracted from 850 μL of buffy coat using a cartridge-based, magnetic bead extraction methodology (Maxwell® 16 Blood DNA Purification Kit, Promega, AS1010X and Maxwell® 16 Research Instrument [Promega, AS200-HS]).
DNA extraction occurs within the pre-filled cartridge (supplied in Promega, AS1010X) which is split into wells containing lysis buffer, MagneSil™ Paramagnetic Particles (PMPs) and wash buffers. Briefly, DNA is moved between wells by the PMPs and the application of a magnetic force to disposable plungers. Cells are lysed by a guanidine-based lysis buffer alongside mechanical lysis from the disposable plungers. DNA adsorbed by silica coated magnetic beads is moved through a series of wash wells by a plunger. Salts from lysis and other impurities that may inhibit downstream processing (e.g. haem, proteins etc.) are removed during the washing. Once purified, DNA is eluted from the beads in Tris-EDTA-based lysis buffer, is assisted by heating at 56 °C.
To maximise the DNA yield and purity from the source material (standard protocol extracts from 250 μL buffy coat), the lysis buffer (Promega, A826E) was increased by 600 μL, the wash buffer (Promega, MD1412) by 1 mL in two wash wells and the cycle through the cartridge was repeated a further two times.
Following extraction, the DNA was aliquotted across three tubes; one for primary storage at UK Biobank (425 μL), one for back-up storage at UK Biobank back-up centre (425 μL) and one for genotyping (50 μL). The primary aliquot was quantified on the Trinean DropSense™ 96.
The quality metrics for a 96-well plate to automatically proceed to genotyping on the Axiom® array was 80% of the plate must have a DNA concentration > 10 ng/μL. Prior to shipping a plate for genotyping, the measured DNA concentration and quality (by absorbance at 260/280) of the stock DNA (sibling to the DNA shipped for genotyping) were assessed from the concentration obtained from the Trinean DropSense™ 96. Results from QC checks were entered into the UK Biobank Laboratory Information Management System (LIMS) and are a data field that can be requested by researchers.
Shipment
Following DNA extraction and quantification, DNA was stored at −80 °C. Plates were sent for genotyping to the Affymetrix Research Services Lab, Santa Clara, CA, USA (ARSL) approximately weekly. Plates were shipped on dry ice and accompanied by an electronic sample manifest containing an anonymised participant identifier plus the gender, ethnicity and geographical location of the participant to which the sample pertains (latter for QC purposes).
Pre-genotyping at affymetrix
Genotyping was performed at the ARSL as per the Manufacturer’s Instructions [13] using the UK Biobank Axiom® Array or UK BiLEVE Axiom® Array.
In summary, samples were thawed and homogenised (incubated at 37 °C for 2 h) prior to PicoGreen® quantification to establish the volume of DNA required for normalisation to a concentration of 10 ng/μL. Plates where < 80% of samples had a concentration of > 10 ng/μL required authorisation before proceeding to genotyping. After normalisation, two controls were added to the plate and samples entered the Axiom® assay workflow [14]. Samples failing initial QC (<95% of markers measured could be confidently genotyped [call rate]) were re-processed. If samples failed re-processing a second sample from the participant was extracted, where available.
Data analysis was performed as per Manufacturer’s Guidelines [15]. Any deviations from standard protocol are documented in [14].
Data QC at WTCHG
QC procedures applied to the UK Biobank genotyped data are described in [16]. Upon completion of QC, data is passed to UK Biobank for release to approved researchers.