Dynamic time warping on sn- and sc-RNA-seq trajectories of Drosophila adult and larvae testis enables contrasting the different germline developmental stages
Authors: Soumitra Pal 1; Amelie Raz 3; Sharvani Mahadevraju 2
Affiliations: 1) National Center for Biotechnology Information, National Library of Medicine, National Institute of Health, Bethesda, MD, 20894, USA; 2) Laboratory of Cellular and Developmental Biology, National Institute of Diabetes and Kidney and Digestive Diseases, National Institutes of Health, Bethesda, MD, 20892, USA; 3) Whitehead Institute for Biomedical Research, Cambridge, MA, USA
Keywords: m. computational models; l. gonads
Capturing the single nuclear and single cellular transcriptomics using snRNA-seq and scRNA-seq have advantages of their own. In snRNA-seq the message made by a cell at a certain time is captured whereas in scRNA-seq, the message stored for a longer period is captured as well. Similarly, RNA-seq profiles (sn or sc) captured from different stages of the life-cycle can reveal differences in the transcriptome across life stages. However, co-analyzing such diverse datasets together to gain biological insights poses significant challenges as the datasets could have batch effects eclipsing the biological differences, or even have significantly different RNA-features. Here we use an alternate approach where we first analyze the datasets separately utilizing then individual nuances and then contrast the analyses by a post-analysis alignment procedure. Specifically, we align the trajectory of germ cells from different data sets and set out to identify if there are any differences in the germ cell developmental stages revealed by the nuclear and whole cell transcriptomics at either life-stages.
We analyzed three datasets from different labs: 1) snRNA-seq on adult testis from Fly Single Cell Atlas, 2) scRNA-seq on adult testis and 3) scRNA-seq on larval testis. The snRNA-seq germline has 21,061 nuclei whereas the adult and lthe arval scRNA-seq have 6,438 and 10,652 cells respectively. The trajectory analyses using monocle3 on these datasets separately arrange the germ cell that progresses from spermatogonia to spermatids via different developmental stages. However, to contrast these individual trajectories for uncovering differences in biology, we adapted Dynamic Time Warping (DTW). We were able to align the three different trajectories on a common pseudotime scale. The alignment of germline adult and larval scRNA-seq pseudotime revealed cell states in adults (elongating spermatids) that are absent in the larvae confirming the legitimate alignment. Buoyed by the accuracy of the alignment by DTW, we next focussed on the difference between snRNA-seq and scRNA-seq of adult germ cells and identified a set of genes for which the cell stops producing the RNAs in the nucleus while differentiating from spermatocyte to spermatid but shows purdurance of transcript in the cytoplasm. While some of these genes are known in literature, some are our novel findings that we set out to validate experimentally.