Minutes of SV skype call on October 13rd

Participants

Victor, Lennart, Najaf, Aby, Leon, Laurent, Abdel

Minutes

Pilot phase SV detection run

• Breakseq: Slavik is working on the reference set using hg19. This seems requiring a long time.
• Breakdancer: Jayne and Joep made some parameter changes on the pipeline installed at Groningen cluster. Will make new SV calls for the pilot dataset.
• Dindel: Laurent finished the SV call and is working now on the filtering part. Aim to deliver the result next week.
• Genomestrip: Bob is continuing on the filtering part. Aim to deliver the result next week.

Pilot phase validation

• Victor explained the design of 384-well plate for pilot phase validation (ref to the attached slides).
• Validation using primer extension is proposed by LUMC (ref to slides). Ion Torrent will be used. We first need to have linker sequence integrate to the primer. Since break points are required here, we will use SV calls made by Pindel for this validation.
• Using primer extension and breakpoint array are experimental. We will see how good the results are and decide how to continue that in the next phase of validation.
• Laurent suggests validation by checking mendelian inheritance given GoNL trio design. A risk here is that we will miss de novo variants if we throw away medelian errors.
• The estimated time to have the pilot phase validation result is 2 weeks.

Full GoNL dataset

• SV calling based on de novo assembly will be part of GoNL SV pipeline.
• GoNL bams need to be transferred to LUMC and Hubrecht.
• 1-2-3 SV/DWAC-seq: Will be ran at hubrecht cluster which has about 100TB storage. Total GoNL bams is 45TB. To avoid having any trouble with storage, Victor proposed a 2-step approach: (1) copy and process suspicious mapped reads and then (2) deal with the full set.