Sorry for the late update. In the past two months, I finished my Ph.D. applications (hope to hear good news in the next two months) and was busy preparing the PyCon Taiwan 2016. Also, a year-long website development finally came to the end.
Now most things are set so I can back to writing my blog.
Since September, there accumulates at least 5 drafts and I don’t know when I can finish them, so I think I have to change my writing strategy. I will first publish things as soon as information collection is done, and deeper reviews will be given in the following posts. Right now I will focus on Bioconductor (and general Bioinformatics topics) and Django.
Bioconductor is indeed a rich resources for R both in terms of data and tools. And I found I have yet spent time seriously understanding the whole ecosystem, which I believe can drastically lighten the loading of daily analysis.
Bioconductor’s website is informative. If you are familar with R, you should already know that in order to understand the usage of a package, one of the best way is to read its vignettes. Packages on Bioconductor generally have vignettes, which is really helpful and the website makes them accessible. On top of that, they have Courses & Conferences and Workflows. The former section collects all conference materials in the past few years, which contains package hands-on, analysis tutorial, and R advanced topics. It’s a hidden gem to me since I have already found numerous materials worth reading only after a glance over it. The latter one should be well-known. It gives examples of typical analysis workflows.
I’m interested in the following topics in Biocondutor:
- Annotation and genome reference (OrgDb, TxDb, OrganismDb, BSgenome)
- Experiment data storage (ExpressionSets)
- Operations on genome (GenomicRanges)
- Genomic data visualization (Gviz, ggbio)
Keywords in Biocondutors for each topic are attached in the parens, mostly being the package name. For each topic, I’ll put the related resources I collected in the following sections.
Before the listing, I found PH525x series maintained by Rafael Irizarry and Michael Love from Harvard serves as a comprehensive entry point for almost every related topic. The site is the accompanied resources for their edX classes. Both of them worth taking a look.
Annotation and Genome Reference
For example, human comes with
Experiment Data Storage
ExpressionSet helps store the expression experiment data, which one can combine expression values and phenotypes of the same sample. Additionally the experiment data (like descriptions of GEO dataset) can be attached as well.
Operations on Genome
I haven’t gone into the details, but operations about genomic ranges are often tricky and more importantly, badly optimized.
Also, their paper, “Software for Computing and Annotating Genomic Ranges”, PLOS One should be another overview source of the package.
foverlapfunction worth the comparison, since I already use it and I know it is blazingly fast.
foverlaphandles the overlapping of integer ranges so it can be applied to genomic operation. Its code is quite complex so its mechanism is still a myth to me. I’d like to see its comparison with using database like SQLite.
Genomic data visualization
Basically I can find two packages:
Don’t know their difference yet. Both of them can produce well-done figures. But I think I have some experience with ggbio, which was a bit tricky to use. So for now I will go for Gviz.
- Visualizing genomic features with the Gviz package given at Bioc Europe 2012 has a decent introduction about Gviz.
- The Gviz User Guide looks very comprehensive, which also cover usage with expression and alignment results.
These resources should be enough for weeks of trying. It’s excited to find so many useful tools.
So, good luck to me for my Ph.D. application, PyCon Taiwan 2016, and a shorter blog posting frequency.