Thursday, February 11, 2016 21:35
Mariateresa de Cesare
Mariateresa de Cesare, University of Oxford “Unlocking the heterogeneity of the human transcriptome utilizing the ONT MinION TM”. The lab has been working with the ONT MinION for about two years. Mariateresa presented some of the Pros of RNA-seq, as well as the Cons, mainly the need to reconstruct transcripts from short reads. Ideally we'd be reading through the whole transcript to get direct counts of isoforms rather than inferring them from short-reads.
Described work on H1 hesc lines on PacBio which found over 200 novel isofroms - in a "well characterised" sample. The Oxford lab wanted to see if they could do even better with the MinION. This should have almost no limit on read-length (the current longest read is well over 100kb) so virtually all human transcripts could be read as single-reads.
Mariateresa described the library prep; generate cDNA using SMARTseq protocol and ONT adapter ligation (leader, hairpin and tethers) that allows concentration of the library at the pore and sequencing of the two strands of the library molecule. Sequencing generates full-length cDNA reads aligned with GMAP with read length distribution peaking at around 500-1000bp but extending past 3000bp. They did find novel transcripts including the previously reported HPAT2.
They are aiming to increase the amount of data, you need low millions of reads to count isoforms accurately. The next version of flowcells should increase read numbers significantly...watch out at London Calling 2.
Joel Malek, Weill Cornell Medical College in Qatar “AVA-Seq: a method for all-versus-all protein interaction mapping using next generation sequencing”. Described the challenges getting to an all-vs-all interaction screening with NGS; high transformation efficiency, get rid of colony picking, etc. NGS allows a statistical cut-off to be applied to the strength of the interaction using a modified Bacteriomatch system from Agilent and in-solution methods avoiding colony-pick.
Israel Steinfeld, Agilent Technologies “Improved methods and analysis tools for efficient CRISPR/Cas genome editing”. Describing TC-RNA technology making RNA synthesis as simple as DNA synth. Can routinely make 120bp RNAs and include modifications and use these to enhance genome editing efficiency (see Nature Biotech article). They are estimating the number of DSB events by counting reads with InDel's around the targeted cut site, execpt to see an accumulation of these and use %InDel as a reporter of efficiency. Showing a modified sgRNA for IL2Rg gene in K562 cells with 81%InDel - a strong target in a susceptible cell line, but small inDels are not easily captured. In the off-target sites they see less than 1-2%InDel. Have designed capture probes sets for pull-down of predicted off-target regions to allow better characterisation of CRISPR edits.
Ji Lee, JGI "Nanopre sequencing at the JGI". Very diverse projects, including the fistulated cow - remember that from about four years ago? Initially tested the Nanopore using standard methods. See very high variability in quality of flowcellls. Yields have increased over time, can get over 500Mb of 2D yield on MAP006 chemistry. They have been working on the low-input protocol using 100ng of DNA, and saw similar results to the standard prep. Now looking at possible applications, focusing on metagenomics using a mini mock metagenome and long-read (20Kb) protocols the first instance.
James Hadfield, University of Cambridge “Progress in developing a nanopore rapid cancer MDX test”. :-(
GiWon Shin, Stanford University “STR-Seq: a massively parallel microsatellite sequencing and genotyping technology”. Targeting over 1000 STRs simultaneously in his method. Three major challenges: 1) random fragmentation of genomics DNA leads to truncation of STRs, fragments break in the STR, making analysis tough, 2) PCR adds artificial InDels, 3) alignment can introduce repeat artefacts. Using CRISPR/Cas9 to target region upstream of STR and single-primer targeting to remove amplification. Targeting up to 2400 STRs with 1700 targeted with guide RNAs. STR-seq has very high correlation to capillary electrophoresis R0.99. Trio analysis found 96% concordance in shared STR calls. Importantly this amplification-free method reduces artefacts over the normal PCR-based methods e.g. stutter peaks, which can be up to 60% of the total data.
Jonathan Reichel, Memorial Sloan Kettering Cancer Center “Non-invasive somatic mutation profiling of liquid biopsies using capture-based next generation”. MSK sequenced 8500 cases for 410 cancer genes to improve diagnosis, improve therapy, recruit into clinical trials. Discussed tumour evolution over time and the problems of sampling from heterogeneous tumours, particularly via a biopsy. Showed a very nice comparison of ddPCR and ctDNA-seq which had very high concordance, and data on CNV in ctDNA (unclear what methods are being used), with very concordant results from tissue when compared to 3 ctDNA samples separated over three months. Also showed CNV in CSF sequencing in leptomeningeal disease (I'd really like to know how they are doing the CNV detection: lab and bioinformatics). They were seeing very low mutant allele frequency, generated 200,000x but theoretical maximum coverage is limited by mass of adapter ligated DNA input, using 10ng ctDNA max is probably only 3000x. Improving results using UMIs (molecular indexes), also trying to improve ctDNA extraction and library prep.
Mariateresa de Cesare, University of Oxford “Unlocking the heterogeneity of the human transcriptome utilizing the ONT MinION TM”. The lab has been working with the ONT MinION for about two years. Mariateresa presented some of the Pros of RNA-seq, as well as the Cons, mainly the need to reconstruct transcripts from short reads. Ideally we'd be reading through the whole transcript to get direct counts of isoforms rather than inferring them from short-reads.
Described work on H1 hesc lines on PacBio which found over 200 novel isofroms - in a "well characterised" sample. The Oxford lab wanted to see if they could do even better with the MinION. This should have almost no limit on read-length (the current longest read is well over 100kb) so virtually all human transcripts could be read as single-reads.
Mariateresa described the library prep; generate cDNA using SMARTseq protocol and ONT adapter ligation (leader, hairpin and tethers) that allows concentration of the library at the pore and sequencing of the two strands of the library molecule. Sequencing generates full-length cDNA reads aligned with GMAP with read length distribution peaking at around 500-1000bp but extending past 3000bp. They did find novel transcripts including the previously reported HPAT2.
They are aiming to increase the amount of data, you need low millions of reads to count isoforms accurately. The next version of flowcells should increase read numbers significantly...watch out at London Calling 2.
Joel Malek, Weill Cornell Medical College in Qatar “AVA-Seq: a method for all-versus-all protein interaction mapping using next generation sequencing”. Described the challenges getting to an all-vs-all interaction screening with NGS; high transformation efficiency, get rid of colony picking, etc. NGS allows a statistical cut-off to be applied to the strength of the interaction using a modified Bacteriomatch system from Agilent and in-solution methods avoiding colony-pick.
Israel Steinfeld, Agilent Technologies “Improved methods and analysis tools for efficient CRISPR/Cas genome editing”. Describing TC-RNA technology making RNA synthesis as simple as DNA synth. Can routinely make 120bp RNAs and include modifications and use these to enhance genome editing efficiency (see Nature Biotech article). They are estimating the number of DSB events by counting reads with InDel's around the targeted cut site, execpt to see an accumulation of these and use %InDel as a reporter of efficiency. Showing a modified sgRNA for IL2Rg gene in K562 cells with 81%InDel - a strong target in a susceptible cell line, but small inDels are not easily captured. In the off-target sites they see less than 1-2%InDel. Have designed capture probes sets for pull-down of predicted off-target regions to allow better characterisation of CRISPR edits.
Ji Lee, JGI "Nanopre sequencing at the JGI". Very diverse projects, including the fistulated cow - remember that from about four years ago? Initially tested the Nanopore using standard methods. See very high variability in quality of flowcellls. Yields have increased over time, can get over 500Mb of 2D yield on MAP006 chemistry. They have been working on the low-input protocol using 100ng of DNA, and saw similar results to the standard prep. Now looking at possible applications, focusing on metagenomics using a mini mock metagenome and long-read (20Kb) protocols the first instance.
GiWon Shin, Stanford University “STR-Seq: a massively parallel microsatellite sequencing and genotyping technology”. Targeting over 1000 STRs simultaneously in his method. Three major challenges: 1) random fragmentation of genomics DNA leads to truncation of STRs, fragments break in the STR, making analysis tough, 2) PCR adds artificial InDels, 3) alignment can introduce repeat artefacts. Using CRISPR/Cas9 to target region upstream of STR and single-primer targeting to remove amplification. Targeting up to 2400 STRs with 1700 targeted with guide RNAs. STR-seq has very high correlation to capillary electrophoresis R0.99. Trio analysis found 96% concordance in shared STR calls. Importantly this amplification-free method reduces artefacts over the normal PCR-based methods e.g. stutter peaks, which can be up to 60% of the total data.
Jonathan Reichel, Memorial Sloan Kettering Cancer Center “Non-invasive somatic mutation profiling of liquid biopsies using capture-based next generation”. MSK sequenced 8500 cases for 410 cancer genes to improve diagnosis, improve therapy, recruit into clinical trials. Discussed tumour evolution over time and the problems of sampling from heterogeneous tumours, particularly via a biopsy. Showed a very nice comparison of ddPCR and ctDNA-seq which had very high concordance, and data on CNV in ctDNA (unclear what methods are being used), with very concordant results from tissue when compared to 3 ctDNA samples separated over three months. Also showed CNV in CSF sequencing in leptomeningeal disease (I'd really like to know how they are doing the CNV detection: lab and bioinformatics). They were seeing very low mutant allele frequency, generated 200,000x but theoretical maximum coverage is limited by mass of adapter ligated DNA input, using 10ng ctDNA max is probably only 3000x. Improving results using UMIs (molecular indexes), also trying to improve ctDNA extraction and library prep.
No comments:
Post a Comment
Note: only a member of this blog may post a comment.