Forum:Any good idea for bioinfo final presentation?

Question

[anran04100]

last semester, we've learn how to analyse date by R on our bioinformatics class.

Our class basically focused on the idea of analysing the data of multi-mics, my classmates have already choosed the topic of PCA, PLS, tSNE and WGCNA. Does anyone could suggest me a good topic according to this? I do really have trouble in choosing things, this presentation is so essential for me. Please guys, I am kind of exhausted.

Thanks!

[russhh]

How to contribute to Bioconductor

[cpad0112]

seems most of them are dimension reducing tools / classifiers to me. You can pick up classifiers such Random forest, CART, SVMs, Lasso regression etc.

Answer 1

[russhh]

This thread may get closed because you have shown relatively little effort.

But before it does, I think it's important that you identify part of bioinformatics, or part of the R/bioconductor ecosystem that is important/interesting to you. It will be extremely hard for you to write an interesting talk on something unless that something sparks an interest for you.

Could you expand upon the purpose of the talk and give an outline of what the expected contents of the talks should be:

• are you to present a data-analysis of your own,

• are you to summarise a data-analysis method,

• are you to able to discuss the software design / architecture underpinning this stuff,

• are you able to discuss the social aspects or the publication aspects of coding and releasing packages.

You should take a step back and think about what the challenges in bioinformatics are in 2020, and (based on your question) how does R solve those problems. For me the biggest challenge is reproducibility.

[anran04100]

First of all, I do appreciate your suggestions.

Since this is my first year of college, I didn't start my own data-analysis. So, I'm going to introduce a type of method to anaylse data. I think I am gonna focus on multi-omics, but there are so many things I don't know that well so it kind of got me there. Maybe I can introduce a package in Bioconductor.

Thanks for your generous ideas.

[Carambakaracho]

I admit, had you asked me the questions above at the level of the OP with the breadth of bioinformatics today, I was overwhelmed, too and I started when Koonin's COGs where bleeding edge...

Answer 2

[Carambakaracho]

I'm not the best source, as I'm behind on most of the active research, but I remember the feeling. For beginners it's not so easy to separate the wheat from the chaff, imo this would be your teacher's responsibility, though.

Let me give it shot: for data analysis milestones comparable with what you mention are

• short read alignment using burrows wheeler transformation, eg. bwa-mem or bowtie2.
• RNAseq analysis: DESeq2. EdgeR, and limma/voom for data analysis, STAR for alignment

None of this is trivial, though, but neither is PCA/PLS/tSNE. Good luck.

[GenoMax]

Carambakaracho : I think OP wants the topic to be related to R since the class is about R.

[Carambakaracho]

bloody **, I should pay attention to details...

Well, technically at least, DESeq2. EdgeR, and limma/voom are R . For what it's worth anyway

[Friederike]

Since it's also about multi-omics, maybe something along the lines of uisng limma for RNA-seq vs. limma for metabolomics data.

That being said, maybe presenting on the Bioconductor philosophy of specialized objects for certain data types might be a good exercise (going from Biostrings to RangedObjects to SummarizedExperiment to SingleCellExperiment and MultiAssayExperiment). The power and usefulness of those objects is often only understood once you have to deal with a package outside the BioC universe (yes, Seurat, looking at you [but not just you]!)

EDIT: just wanted to provide a couple of pointers about the different object types:

• http://waldronlab.io/MultiAssayExperiment/, https://cancerres.aacrjournals.org/content/77/21/e39
• Orchestrating high-throughput genomic analysis with Bioconductor
• Using R and Bioconductor in Clinical Genomics and Transcriptomics