Maca

Following a removal of linker/primer contaminations and artificially duplicated reads, the remaining 477,699 reads (average length of 206 nucleotides) were used for repeat analysis. The analysis was performed as described by Macas et al. (2007) [15 (link)], employing TGICL [47 (link)] and a set of custom-made BioPerl scripts for similarity-based clustering and assembly of reads. The clustering parameters used by a tclust program (part of TGICL) were set to consider pairwise similarity of two reads significant if it involved an overlap of at least 150 nucleotides with 90% or better similarity, representing at least 55% and 70% of the length of longer and shorter read respectively (OVL = 150 PID = 90 LCOV = 55 SCOV = 70). The reads within individual clusters were assembled into contigs using TGICL run with the -O '-p 80 -o 40' parameters, specifying overlap percent identity and minimal length cutoff for cap3 assembler. Repeat type identification was done using blastn and blastx [48 (link)] sequence-similarity searches of assembled contigs against GenBank, and by detection of conserved protein domains, using RPS-BLAST [49 (link)]. Tandem repeats within contig sequences were identified using dotter [50 (link)]. The classification of LTR retrotransposons into distinct lineages and clades was done using phylogenetic analyses of their RT sequences [15 (link)]. Alignment of RT sequences was carried out with ClustalX [51 (link)] and the phylogenetic trees were calculated using neighbour-joining method. The trees were drawn and edited using the FigTree program.
Microsatellite sequences were identified using Tandem Repeats Finder [52 (link)] and TRAP [53 (link)] programs, while a BioPerl script was used to identify ISBP loci [54 (link)]. Identification and classification of repetitive sequences within BAC clones was done via PROFREP web server [45 ] utilizing repeat-specific databases of 454 reads prepared in this study. The server performs BLAST-based searches against databases of whole-genome or repeat-specific 454 reads and generates plots of similarity hits along the query sequence (number of hits is proportional to copy number of the query in the genome).

Prism 8 software (GraphPad Software Inc., San Diego, CA, USA) was used for statistical analysis. The data are shown as mean ± SEM. Unpaired or paired Student’s t-tests were used to analyze between groups or in the social interaction test, respectively. To analyze the effects of groups (control or VPA) × treatment (water or maca), two-way analysis of variance (ANOVA) was used. Two-way repeated measures ANOVA was used for either the effects of groups (control or VPA) × trials or groups (control–water, control–maca, VPA–water, VPA–maca) × trials, followed by post hoc analysis with Tukey’s multiple comparison test.

Once cell-type markers are identified, Mapping cell-type labels to query data is performed using MACA^{20 (link)}. Briefly, for each cell, MACA generates two labels - a per-cell cell-type Label 1 and group-based clustering Label 2. Then, MACA maps clustering Label 2 to cell-type Label 1 to get the overall cell-type annotation. In our previous study, we used different clustering parameters to generate multiple Label 2s, for the purpose of reproducibility^{20 (link)}. In this study, we also ran Louvain community detection with a range of clustering parameters to get multiple clustering Label 2s. These include clustering resolution 3, 5, 7 with 5, 10, 15 as neighborhood sizes to over-cluster cells. With multiple clustering Label 2s, we were able to map them to Label 1 and get a more reproducible ensembled cell-type annotation. To accommodate for large-scale scRNA-seq data, we split the whole data into N batches and ran MACA with one batch per CPU core.