[Geuvadis_rna_analysis] Geuvadis RNAseq analysis group TC minutes

Hello all, A brief summary of the first analysis group TC. Action items in *bold*. Next call Thursday January 12, 2pm. Attending: Tuuli, Micha, Natalja, Marc, Mathias, Tim, Olof, Marta 1) Kit situation - Geneva and Barcelona have both mRNA and miRNA kits - Berlin and Kiel have miRNA kits - Munich, Uppsala and Leiden (?) don't have any kits - *All the labs* should send Tuuli an update before Christmas indicating if they have received the kits or not. We'll wait until early January and see then if we need a plan B. - the labs that have kits are on schedule with the sequencing 2) Low-level data processing (mapping, quantification etc.) - *Tuuli* will upload the fastq files and bams (from bwa) from their mRNA seq by the end of the week. * - Tuuli* will define a sandbox dataset of 24 and 5 samples from UNIGE for testing purposes. - *Micha* will find out when GEM is likely to be published - We shouldn't spend too much time figuring out how to map the reads - this has been done already. However, there are a couple of things that we should test: - *Marc* will analyze the level of genomic contamination in the 24 samples - Tuuli has concerns about reference allele mapping bias affecting quantifications. *Natalja* will run Tophat for 5 samples with the normal reference (hg19) and a reference masked for all common 1000g variants (Tuuli will provide this) to see if masking leads to a big loss in mapping. *Micha *will check whether it's possible with GEM. If masking is not feasible, we can consider other options for dealing with this bias... The following was planned for the final dataset: - Micha will run his whole pipeline (GEM for mapping, SNAPE for variant calling, Flux for deconvolution and normalization, AStalavista for alternative splicing analyzes..). - Natalja will run the EBI pipeline using bwa and/or tophat, and quantify exon counts and/or RPKMs. bwa+exon counts would enable direct comparison with earlier eQTL results from Manolis's lab - regarding normalization, both PCA and specific covariate based approaches are possible - we'll have to see what the data looks like - all analyses should be run with duplicates - Geneva has seen that in RNAseq data duplicates rarely seem to be PCR artefacts 3) FTP instructions - *Natalja* will send an email to the analysis group about the FTP instructions that are now on the ftp site I hope I remembered at least the most important things! I'll let you know when the data is one the ftp site. best regards, Tuuli -- Tuuli Lappalainen, PhD Department of Genetic Medicine and Development University of Geneva Medical School CMU / Rue Michel-Servet 1 1211 Geneva 4 Switzerland Tel. +41-(0)22-3795550 tuuli.lappalainen(a)unige.ch