Dear readers, two weeks back we asked for your suggestions for best bioinformatics innovations of 2012. In addition to topics covered in our blog over the entire year, we received several other good suggestions by email and in the comment section. We also went through all 2012 issues of nine major journals, read the abstracts and shortlisted few more papers that seemed interesting. Additionally, we checked all 2012 posts of four bioinformatics-related blogs we follow to make sure nothing is missed. We could not read everything submitted to arxiv.org, but Haldane’s Sieve came to be of great help to reduce our load.
That was lot of work, but then we faced the dilemma of how to judge between topics as diverse as a very good alignment algorithm and an excellent educational resource in bioinformatics. They solve problems so different that it is nearly impossible to say which one is ‘better’. Should we rank them all by some metrics and pick top five, to end up in a situation, where all top innovations were assembly algorithms? Or should we try to select one each from various categories? We went for the later, but also decided to disclose the categories so that there is no confusion among our readers.
We will list our final results in five categories – (i) best blog/social media+twitter feed, (ii) most innovative assembly algorithm, (iii) best educational resource, (iv) best scientific contribution related to bioinformatics and (v) best bioinformatics journal. For entertainment, we also include the most over-rated scientific discovery of 2012 (thankfully happens to be not bioinformatics-related).
Please feel free to discuss in the comment section, if you do not agree with our choices or would like to suggest other interesting contributions missed by us.
Best blog + Twitter feed – Getting Genetics Done + @genetics_blog
Blogging is a new platform for communication, and this method started to gain traction among scientists over the last two years. Most online-savvy researchers also maintain a Twitter feed to post other interesting information that may or may not be worthy of a full blog post. Additionally, online forums like SeqAnswers have become invaluable tools for bioinformatics researchers, because they bring many practitioners together and allow them to share ideas. It is very difficult to rank various social media channels, because all channels have their own whims and personalities. Still, we decided to create a shortlist from a number of excellent resources listed here, and asked the question – which channel, if it goes away, would be the one most missed by us? For bloggers also maintaining Twitter feed, we combined the contributions in those two channels together.
Following blogs, Twitter feeds and social media channels were shortlisted (not ranked in any order).
1. Openhelix, Twitter feed @OpenHelix
3. Titus Brown’s blog Living in an Ivory Basement and Twitter Feed @ctitusbrown
4. Getting Genetics Done + @genetics_blog
5. SeqAnswers online forum.
6. Jonathan Eisen’s The Tree of Life
All seemed of nearly equal importance to the community and to us. Still, we felt that the daily update of important papers by @genetics_blog saved us tons of time, and were the most invaluable resources.
Best Assembly Algorithm – Minia
The focus of our blog was to cover various NGS-related assembly algorithms, and therefore we decided to make it a different category. We shortlisted the following innovations (not ranked in any order).
They were all so good and addressed such widely different aspects of genome assembly that it was very difficult to compare them against each other. SGA used a new graph technique to solve contig construction part of genome assembly. Titus Brown’s paper worked on data reduction, which was essential for a large assembly. Cortex addressed haplotype difference issue, and rectangular graphs were proposed to do scaffolding very elegantly.
We could not do proper justice to Ray Meta, because the paper came out very late in the year and we did not have enough time to fully grasp all details like we did with other ones.
We selected Minia for two reasons.
(i) Firstly, the algorithm proposed an elegant way to reduce redundancies in de Bruijn graphs. When a de Bruijn graph is constructed, each k-mer has extensive overlap with the consecutive k-mer, and longer the k-mer size is for graph construction, more is the redundancy. Reducing the graph size and therefore the needed RAM was a very important issue for handling very large assembly problems, and Minia provides a method that leads to dramatic data reduction without leading to any approximation in the assembly.
(ii) The optimal bits/k-mer analysis of NGS data (Fig. 2) presented in the Minia paper was very interesting, and may be helpful for other papers working on de Bruijn graphs.
Nevertheless, other assemblers mentioned above appear so innovative that we constructed wiki pages for each of them to analyze and understand the codes. The authors have been very generous in making their codes open-source, and we should make every effort to make best use of that opportunity, and not consider each bioinformatics program a ‘package with parameters and specifications’.
Best Educational Resource – Rosalind
We considered following alternatives -
Rosalind beats them all hands down. The topic on Rosalind was also very popular in Reddit.
Best Scientific Contribution related to Bioinformatics
A pair of papers published by Joanna I. Sułkowska et al. in June 2012 issue of PNAS got our top vote.
Conservation of complex knotting and slipknotting patterns in proteins
Genomics-aided structure prediction
The above decision will take more space to explain, and we will do that in a subsequent commentary. Following papers or groups of papers were considered in our final list.
1. BLASR paper
4. SPAdes paper trio
Pathset Graphs: A Novel Approach for Comprehensive Utilization of Paired Reads in Genome Assembly – Son K. Pham, Dmitry Antipov, Alexander Sirotkin, Glenn Tesler, Pavel A. Pevzner, Max A. Alekseyev
SEQuel: improving the accuracy of genome assemblies – R. Ronen, C. Boucher, H. Chitsaz, P. Pevzner, Bioinformatics. Jun 2012
Paired de Bruijn Graphs: A Novel Approach for Incorporating Mate Pair Information into Genome Assemblers -Paul Medvedev, Son Pham, Mark Chaisson, Glenn Tesler, Pavel Pevzner
5. Protein Folding papers
Best Bioinformatics Journal – J. Computational Biology
We must admit that we started with a bias, and thought we could pick ‘arxiv.org’ as the most exciting up-and-coming bioinformatics journal. That was in line with many of our earlier commentaries on open publication and internet communication. The last thing we thought we would do was to pick a closed journal.
Then we decided to compare our choice with various journals we check from time to time, and the quality of unusual bioinformatics-related papers published last year in J. Comp. Biol. blew our mind. They were far more thoughtful articles than merely combining various computational techniques to develop ‘tools’ or ‘databases’. We like to go through many of the published articles as time permits, and discuss in our blog.
Most Over-rated Scientific Discovery – Discovery of Higgs Boson
Discovery of Higgs boson and killing of Osama bin Laden were two stories over-hyped by mainstream media in 2012, and they have many similarities.
1) They were both very expensive ‘needle-in-a-haystack’ operations.
2) Both events were deemed important in an earlier era, but the world changed so much since then that their relative significances went down drastically.
Here were our thoughts on discovery of Higgs boson, when the news was announced.