Friday, February 12, 2016 21:10: I'm flitting between sessions tonight...first up is a talk by a users of my lab, we generated the data for several hundred C.eleganss genomes using Nextera and HiSeq 2500 - not quite a 1000 genome project yet!
Geoffrey Nelson, MRC Laboratory of Molecular Biology “A high-throughput genomics approach to efficiently link genes to phenotype”. The De Bono lab uses C.elegans as a model for how nervous systems coordinate behaviour, it's a good model as it has 20,000 genes but in a genome ust 100Mb in size.
How to efficiently pinpoint causal mutations from a forward genetics screen: usually use a mutagen to induces point mutations, cross phenotype -ve with phenotype +ve, generate around 50 recombinants, select F1 phenotype -ve and sequence. Have used a simplified screen which remove the need for crossing. They are going from mutagenesis to phenotype -ve and directly into WGS. EMS and sequencing of 450 strains. Selection for EMS mutations identifies mostly missense mutations, about 32 mutations per strain. Are able to pinpoint a causal mutation in around 250 of 450 strains. Remaining mutants can be characterised by low-depth mapping-by-sequencing, as low as 5x coverage of the C.elegans genome.
Geoff Otto, Foundation Medicine “Assessment of the relative clinical utility of ctDNA and tissue biopsies for the detection of actionable genomic alterations in routine clinical oncology specimens”. Foundation Medicine has been built on the analysis of FFPE sections, but liquid biopsies rock and Foundation have now jumped onto the bandwagon. Geoff described the 5 critical features that FM consider important: 1) it must be comprehensive, 2) it must be of the highest quality (Geoff discussed performance metrics and controls), 3) it must be accurate at low allele frequency with high specificity, 4) needs to be validated (e.g. CLIA), 5) needs to be operationally useful i.e. fast TAT with clinically relevant samples.
Aiming to get to under 1% MAF. 10-20ml blood or 5-10ml plasma, at least 25ng of extracted ctDNA and use a smear analysis to quantify ctDNA content. Spent lots of time maximising the library prep using molecular and sample barcodes (collaboration with NEB). The assay achieves 50-70% overall conversion efficiency of ctDNA to NGS library. Deep coverage is very uniform at 5000x greater than 99% are at least 2500x - this is high-coverage sequencing. FM spent a lot of time considering contamination issues and built baiting strategies into their pipelines to detect this, they are assuming nothing is perfect and want to know when it has gone wrong. Wonderful obs/exp allele freq. Observed 100% concordance with FoundationOne or ddPCR. Also looking at CNV in ctDNA showed ERBB2 amplification, but this is the easiest case to detect, many smaller or lower-amplified events will be tougher.
Geoff then showed some preliminary clinical results, but pointed out that the initial results said a lot about sample quality, many were limited or low-quality plasma. Pointed out the challenges of poorly stored prepped plasma - particularly dilution of ctDNA with gDNA from lysed PBMCs. Streck tube collected DNA was used as a comparator and looked great and this is what they are recommending.
Nice work, and coming out in a peer-reviewed journal near you soon!
Andrew Mungall, BC Cancer Agency Genome Sciences Centre “Detection of genomic rearrangements in archival lymphoma tissues using targeted capture sequencing”. Recurrent translocations are a hallmark of B cell lymphomas e.g. t(14;18), t(3;14) BCL2, BCL6. Andy focused on MYC rearrangements as of particular importance which give very poor survival, particularly if you have "double hit" disease (BLC2 + MYC). Novel rearrangements have been found but the large repository of FFPE tumours has not been investigated for break point discovery. The study Andy is describing tonight aims to use targeted sequencing of FFPE tumours to develop an optimised rearrangement detection workflow and discover novel rearrangement partners. Showed examples for a recent J.Path paper.
They have automated the extraction of FPE using Agencourt FormaPure extraction from 120mm sq sections (1 to 5 10um scrolls) with yields from 200ng to 7ug as determined by Quant-IT assay! Are using a plate-based library prep from 100ng with Covaris shearing, double-bead SPRI cleanup to remove smallest fragments, NEB PreCR and then a standard library prep with 4-6 cycles of amplification. Capture is of 7.8 Mb with Agilent probes in a pooled capture reaction, sequencing on HiSeq 2500 V4.
They have run 157 cases in the pilot and all were previously run on a FISH break-apart assay, which had already identified rearrangements. They saw very low %on target or %useable bases (as bad as 15%), but a mean target coverage of 300-1000x. For MYC 78% of NGS results matched the FISH break-apart assay.
Again another nice piece of work coming out in a peer-reviewed journal near you soon (currently responding to reviewers comments).