Venue: Biomedicum 1, Meeting room 8-9, P-floor

Timing: 15:30 onwards

Presenter: Mehreen Ali, PhD student, FIMM

Missing data imputation methods in proteomic datasets

Mass spectrometry (MS)-based proteomic profiling has the potential to study comprehensive protein profiling of biological samples and thus has shifted the focus of research from qualitative to quantitative analyses. MS-based proteomics provide opportunity for more global profiling of post-translational modifications, in terms of yielding proteome-wide information about cancer cell signaling activity that is not accessible by genomics or transcriptomics alone. 

However, a substantial amount of data is missing at peptide/protein level primarily due to low-abundance peptides and/or poor ionization. Missing values in proteomics datasets limit the information extraction from proteomics datasets using statistical and machine learning methods, and thus have detrimental effect on downstream analyses.

In this talk, I will give a general overview of imputation methods adapted for proteomics datasets

Tentative schedule for Fall session 2017

Venue: Biomedicum 1, Meeting room 3, P-floor

Timing: 15:30 onwards

Presenter: Abhishekh Gupta, Post-doc, Group Aittokallio, FIMM

Drug response quantification: past, present and future

Venue: Biomedicum 1, Meeting room 5-6, 3rd Floor

Timing: 16:00 onwards

Presenter: Alok Jaiswal, PhD student, FIMM

Methods for analyzing genome-wide loss-of-function screening data

Venue: Biomedicum 1, Meeting room 8-9, P floor

Timing: 15:30 onwards

Presenter: Dr. Ari Löytynoja, Group Leader, Institute of Biotechnology

Generalised model of local template switch mutations

Replication-related template switch mutations are known to create short,
perfect inverted repeats in bacteria. We developed a generalised model
(the four point mutation model; 4PM) that describes local template
switches in DNA replication. Using this model, we studied the mutation
clusters in human and primate sequence data and found that large numbers
of these are caused by template switch events. The 1000 Genomes data
show that 4PM events segregate in human populations and the Icelandic
trio data contain de novo 4PM events. I describe the 4PM alignment
algorithm and the computational analyses involved in the study
(http://biorxiv.org/content/early/2016/09/27/038380). I will then
discuss what the findings mean for reference-based sequencing studies
and for the evolution of genomes.

Venue: Biomedicum 1, Meeting room 5-6, 3rd Floor

Timing: 15:30 onwards

Presenter: Dr. Matti Kankainen, Bioinformatics Team, FIMM

Best practices in RNA-seq data analysis

RNA sequencing (RNA-seq) has become an indispensable tool to characterize transcriptomes and understand the molecular indicators of disease and treatment. It is a powerful technology to study gene expression and discover novel isoforms, in addition to which RNA-seq has proven to be highly useful technique to discover expressed coding variants, deep intronic mutations, and fusion-transcripts. In this talk, I will describe how we at the FIMM sequencing center process RNA-seq data from model organisms and will give a sneak preview to the upcoming new analysis features we are developing at the moment.

Tentative schedule for Spring session 2017

Venue: Biomedicum 1, Seminar room 1-2

Timing: 15:30 onwards

Presenter: Boris Vassilev, Ikonen Lab, Department of Anatomy, Biomedicum Helsinki

Document, automate, and share your data analysis with Lir

Lir is a software tool for documenting, automating, and sharing computational work. We have demonstrated that Lir can improve the reproducibility of a complex data analysis. With Lir, it takes less effort to document the analysis, while the generation of all results is automated. Importantly, Lir supports the use of multiple programming languages and approaches in the same data analysis. This allows the user to employ whatever tool is best suited for each step of the analysis

See more details from the publication: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0164023“

Venue: Biomedicum 1, Seminar room 1-2

Timing: 15:30 onwards

Presenter: Teemu Kivioja, Taipale Lab, Genome-Scale Biology Program

Utilizing integer linear programming in experiment design and analysis

Many tasks in bioinformatics involve choosing the best combination
from a vast number of different alternatives, for example the optimal
set of molecular probes from a large number of candidate probes. Many
such discrete optimization problems can be conveniently expressed in
Integer Linear Programming (ILP) or more generally Mixed Integer
Linear Programming (MILP) framework. The improvement of general ILP
solvers together with the increase in computing power has made it
possible to optimally solve many problem instances of small to medium
sizes without developing a task-specific algorithm or resorting to
heuristics. I will introduce how one can use ILP to solve experiment
design and analysis problems through simple examples taken from my own
research including molecular barcode design, CRISPR/Cas9 guide RNA
selection, and selecting a representative set of transcription factor
binding motifs.