In an earlier post, I wrote about five open problems in bioinformatics. In the next several posts, I will select each of them and discuss in some detail. The current post is on the shotgun development biology experiments and related challenges.
I find the development of multicellular organisms fascinating. Usually we consider the entire organism (man or dog or cockroach) as the living entity. When we look closer, we find the organism to be composed of billions of cells, and each of those cells is fully living. Moreover, the entire human being is formed through cell division starting with a single cell, and every cell in the body knows the chemical program to form an entire person. This was best demonstrated through the work of Shinya Yamanaka, who in 2006, changed the state of skin cells of mice to the earliest (or pluripotent) stem cells by adding a set of chemicals. That means each of those reprogrammed cells could then divide to form a whole mouse.
The process of division of embryo, starting with a single cell to a fully structured body, is highly deterministic. In 1977, John Sulston and Robert Horvitz published the cell-lineage map of C. elegans, where they tracked down the step-by-step cell divison of every single cell to form about 1000-cell adult body. Brenner linked that cell division process to the expression of genes within the cells. Here is an excellent review covering 40 years of research on that topic.
How do we replicate that research in mammals, where trillions of cells need to be tracked instead of only 1,000 cells? Ideally we need gene expression from every single cell, and that is where single cell RNAseq experiments come in. Also, we need a method to track down the division of each cell, and that is where the process gets complicated. Do we place cameras to record cell divisions and hire a million graduate students to monitor every step? In the context of human genome sequencing, that is what the smartest people in the room proposed in their clone-by-clone method (also check “Against a Whole-Genome Shotgun”).
The shotgun approach means tracking down the gene expression of each cell, or rather a randomly selected subset of cells, at each stage of the developing embryo. Then let computer programs connect the dots. Just like genome assembly, the computational researchers need to convince that this method can solve previous-solved problems. Here is recent paper that reconstructs the development of C. elegans using the described method.
In the following post, I will talk about application of the same tools to understand the development of mammals, and the new statistical algorithms like tSNE and UMAP being used to analyze the data.