Francois Sabot
LightNews LightNews
banner
francoissabot.bsky.social
Francois Sabot
@francoissabot.bsky.social
410 followers 510 following 660 posts
Scientist, biologist, #evolution and #pangenomics, #bioinformatics, #AI and #OpenScience @ird_fr - and a little bit a #geek too
mastodon: @[email protected]
Posts Replies Media Videos
francoissabot.bsky.social
In conclusion, GraTools is a powerful too for a one-tool manipulation of GFA pangenome variation graphs!

It is a member of the GraSuite, feel free to discover !

forge.ird.fr/diade/GraSuite
The GraSuite Logo a suite and a tie showing a graph, with a double strand DNA symbolized on the right
December 3, 2025 at 8:35 AM
francoissabot.bsky.social
You can also extract nodes and paths shared ONLY by a subset of embedded individuals, to chek differences between groups for instance
Venn diagram representing the indica and japonica subspecies specific nodes in the Asian Rice graph from Marthe et al, 2025, computed using specific groups samples. For each subspecies (purple for the indica, and orange for the japonica), we indicates the sum in bp of the total nodes (specific or shared).
December 3, 2025 at 8:35 AM
francoissabot.bsky.social
Subgraph extraction, based on these coordinates (from your favorite locus in your favorite sample, for instance), can be then put into Bandage
Bandage representation of the Sub1 locus in Asian rice graph from [32].The 3 genes of the locus, namely Sub1C, Sub 1B and Sub1A are depicted in yellow, green and red, respectively on the graph structure
December 3, 2025 at 8:35 AM
francoissabot.bsky.social
It is structured around a few key classes and features that interact, after converting the graph in BED+ BAM. It allows to interrogate it based on genomic coordinates from *ANY* embedded genome
Overview of GraTools. The central Gratools class coordinates the interactions with three core classes: GFA, GratoolsBam, and SubGraph. The functionalities and commands are organized into four categories: GFA indexing (index), content information (stats, list samples, list chr), data extraction (extract sub graph, get fasta), and content analysis (core dispensable ratio, depth nodes stat, specific groups sample, get segments by depth). The GFA files serve as input, and outputs are provided in BED, BAM, FASTA, GFA and CSV formats, enabling a comprehensive analysis of pangenome variation graphs.
December 3, 2025 at 8:35 AM
francoissabot.bsky.social
This tool, available as a GNU GPLv3 one, is available here gratools.readthedocs.io/en/latest/

It can extract stats, subgraphs, fasta, count specific nodes and identify specific paths, in an easy and visual way
Screen shot of the readthedocs page saying " GraTools offered a set of commands for analyzing and handling pangenome graphs quickly, whatever the model and therefore the size and the complexity of the GFA file. For example, GraTools allows to extract sequences or subgraphs, convert graphs into FASTA sequences, and identify shared and specific nodes across samples. Additionally, it provides essential statistics for chromosomes and samples data. With multi-threading support and flexible logging capabilities, GraTools is an efficient solution for pangenome graph analysis. GraTools has been developed by IRD and CIRAD, as part of a collaboration between the PANEEC/DIADE and PHIM research teams."
December 3, 2025 at 8:35 AM
francoissabot.bsky.social
We are pleased to release today GraTools, an user firendly tools to manipulate the #pangenome variation #graph in GFA format.

@ird-fr.bsky.social @cirad.bsky.social @inrae-france.bsky.social
Example of statistics obtained using the stats command on the NewRiceGraph MGC.gfa.gz graph: gratools stats -g NewRiceGraph MGC.gfa.gz A nice tabular format in rendered in the terminal, with path names, number of nodes, size, etc...
December 3, 2025 at 8:35 AM
francoissabot.bsky.social
With many colleagues, we tried to gather the future prospects and challenges for #Pangenomics and #Pangenome research in agronomy, that can however be adopted more largely

hal.science/hal-05357866/

@ird-fr.bsky.social @inrae-france.bsky.social #FAIR #OpenScience #vizu
A possible pipeline for the uses of a pangenome graph. This pipeline can be roughly divided into three main sections (on the right): build a graph, using it, and adding reads. A first pangenome graph is created, usually with almost complete haplotype as- semblies. Since graph creating methods vary, and may be fraught with errors (unaligned regions, broken paths), assessing the quality of a graph is useful (this can be applied after each subsequent step). Similarly, it is also crucial to be able to visualize the graph, espe- cially on particular loci of interest (can also be done throughout the pipe-line). The graph can be augmented with additional data increasing allele counts, including known con- tig alignment or read mapping. Several augmentation iterations, with different types of data, can be considered. Since most analyses focus on genes and transposable elements, it is usually useful to add an annotation (available on one or several reference genomes), which projected to the graph. This graph, possibly extended with new variants and an- notation, should be then shared with the community, following FAIR practices. Most analyses then focus on graphs with only one species, and the graph can be analyzed in order to find regions of interest. However, dual pangenomics, which studies the joint diversities of two species, can be envisioned. More generally, many species can also be modeled, for instance in the case of meta-genomics. Low coverage sequencing data of individuals can be compared or aligned to the graph in order to produce large genotyp- ing sets enabling population genetics studies (i.e. coalescence, association). Other omics (e.g. RNA-Seq, ChIP-Seq, Hi-C, LC-MS, BS-Seq etc.) data can be added to the graph, and potentially aggregated to multi-omic layers.
November 25, 2025 at 9:19 AM
francoissabot.bsky.social
Animals are just a few part... from the same location
Visualizing biomass on earth, not only animals but all kingdoms. Mostly plants, then bacteria, fungi, archae, protists, animals, then viruses. Humans are less than 0.01% of the total biomass...
November 12, 2025 at 9:34 AM
francoissabot.bsky.social
Nina compared it to state-of-the-art tools for genome-to-genome transfer (LiftOff), genome-to-graph (VG and ODGI), and GrAnnoT is more conservative and fast, and is the only tool able to go efficiently from genome-to-graph AND from graph-to-genome!
Genome to genome transfer comparison. GrAnnoT and Liftoff run time for 1-12 transfers were measured using the command /usr/bin/time. Liftoff was run both in GFF and DB mode. GrAnnoT values are presented including or not the graph building time.
October 30, 2025 at 3:46 PM
francoissabot.bsky.social
Basically, it converts/transfers coordinates from any embedded genome in the graph to the graph, and from the graph to any other embedded genome! It can transfer with a certain level (user defined) of identity/length, and provide alignments between source and target.
GrAnnoT overview. Step 1: the position of the feature is projected from the source genome to the graph using the positions of the nodes on the source genome. Step 2: the position of the feature is projected from the graph to the target genome (a). The first and last nodes from the feature that are on the target genome (the blue ones) are the ends of the feature in this genome, and everything in between is considered as part of the feature. The differences between the two genomes in this region in terms of path in the graph mirror the differences between the two versions of the feature. If these differences are too important and the transfer does not reach the coverage and/or sequence identity thresholds, the annotation is not transferred (b).
October 30, 2025 at 3:46 PM
francoissabot.bsky.social
Finally, B. prasinos is also known to have a very strange SOC (small outlier chromosome, the 19th), supposedly linked to viral resistance. Using our high quality assemblies, we found some shared sequences at an unexepected length !
Chromosome 19 SOC shared sequences between B. prasinos strains. Chord plot representation of chromosome 19 shared sequences between strains of the geographically close water basins (Left) and between all 24 B. prasinos sequenced strains (Right). Chromosome color corresponds to basin and geographical distribution of strains as shown in Figure 5 : Blue, Arctic ocean ; Green, North sea and English channel ; Red, Mediterranean sea. Only shared sequences longer than 500 bp are displayed.
October 30, 2025 at 3:39 PM
francoissabot.bsky.social
B. prasinos possesses two haplotypes on its "BOC" (Big outlier Chromosome), haplotypes shared between many strains and almost with no variation within. The ratio between the two is basin dependent
Phylogenomic of B. prasinos linked to strain origin and Chromosome 14 BOC haplotypes. (Left) Maximum-likelihood phylogenetic tree of B. prasinos strains based on 1,200 shared genes. (Right) Maximum-likelihood phylogenetic tree of Bathycoccus strains based on 69 shared genes located on chromosome 14. Only branches with bootstrap values higher than 90% are displayed. Gene concordance factors are displayed for all visible branches. Genomic diversity and haplotypes biogeography of Bathycoccus prasinos big outlier chromosome. (A) Alignment of Chromosome 14 from B. prasinos strains RCC4222 and A8. Colors indicate syntenic regions and structural rearrangements. Y axis displays the GC content of the corresponding sequences. (B) Divergences between local single nucleotide polymorphism (SNP) density and genome wide SNP density along a 20 kb sliding window. (C) Divergences between local structural variation (SV) density and genome wide SV density along a 20 kb sliding window. SNP and SV calling was computed for each BOC haplotype against strain RCC4222 for BOC A and strain A8 for BOC B. Characterization of BOC haplotype geographical distribution. (A) BOC haplotypes ratio based on read mapping from TARA ocean metagenomic datasets in stations with B. prasinos relative abundance > 0.1%. (B) BOC haplotype ratio among strains isolated in the Banyuls Bay (Left)(n=55) and the Baffin Sea (Right)(n=183).
October 30, 2025 at 3:39 PM
francoissabot.bsky.social
MAG are really valuable in algae (meta)genomics for species discovery. However, they are not suitable for intraspecific variations: chimerical contigs, fragmented and partial genomes, and so on.
Qualitative comparison between de novo genome assembly from whole-genome sequencing data (WGS) and metagenome assembled genomes (MAGs) of Bathycoccus prasinos. (A) Maximum-likelihood phylogenetic tree of B. prasinos available genomes based on 265 shared genes. Only branches with bootstrap values higher than 90% are displayed. Gene concordance factors are displayed for all visible branches. (B to E) Distribution of principal assembly statistics between available MAGs (n=11) and whole genome assemblies produced from isolated strains of B. prasinos (n=24). Median values are displayed for reference.
October 30, 2025 at 3:39 PM
francoissabot.bsky.social
Surprise ! Using phylogenomics we identified/confirmed the presence of a third species, Bathycoccus catiminus! It was previously supposed to be a subgroup in the prasinos species (B3 type), but is in fact a single species!
Phylogenomic of the Bathycoccus genus and B. catiminus characterization. (A) Maximum-likelihood phylogenetic tree of Bathycoccus strains based on 1,200 shared genes. Species separation is derived from the branch length and marked by colors. The tree includes strain UST710 isolated and sequenced by Xu et al. (2024) for reference. O. tauri and M. pusilla are used as outgroups. Only branches with bootstrap values higher than 90% are displayed. Gene concordance factors are displayed for all visible branches. Transmission electron microscopy images of (B) B. prasinos, (C) B. catiminus strain isolated from the Banyuls bay, and (D) a close up of detached spider web scales from B. catiminus. N: nucleus, Cp: chloroplast, St: starch granule, Sc: scales. Bar = 100 nm. (E) Presence of B. catiminus in Banyuls bay seawater samples from 2019-2020 and 2020-2021 phytoplanktons blooms, identified using specific primers designed from non-conserved sequences of TOC1 ORF specific to B. catiminus. Fig S1: (A) Maximum-likelihood phylogeny and (B) 2D centroid structure of available Bathycoccus 18S rRNA ITS2 region. Only bootstrap values higher than 90% are displayed. Notable structural variations compared to B. prasinos 18S ITS2 region are marked by arrow and circles. 18S sequences were obtained from B. prasinos RCC1105 reference strain, B. calidus RCC716 strain and Bathycoccus sp. G8 strain respectively. Bathycoccus Kara environmental sequence was obtained from Belevich et al. (2021).
October 30, 2025 at 3:39 PM
francoissabot.bsky.social
The assemblies were of very high quality, almost complete, and better than the published reference genome.
Genome assemblies statistics and quality assessment. (A) BUSCO completion score of genome assemblies for the 28 de novo assembled Bathycoccus genomes using the chlorophyta_odb10 library. RCC1105 corresponds to the reference genome produced by Moreau et al. (2012), while RCC4222 corresponds to the de novo assembly obtained from a clonal strain of RCC1105. (B) PHRED quality score of genome assemblies per chromosome. (C) Distribution of principal assembly statistics. The N50 corresponds to the sequence length of the shortest contig at 50% of the total assembly length. (D) Comparison of BUSCO single-copy completion score using complete genome sequence or annotated CDS sequences as input.
October 30, 2025 at 3:39 PM
francoissabot.bsky.social
From a large sampling we obtained 28 high quality genomes for the species
Sampling, sequencing, genome assembly and annotation strategy of Bathycoccus. (A) Geographical distribution of the sampling sites for the selected Bathycoccus strains. The number of strains sampled at each site is indicated. Colors correspond to the oceanic basin of origin: blue, Arctic Ocean; green, Atlantic Ocean; red, Mediterranean Sea; grey, Indian Ocean. (B) Summary of the sequencing, assembly and annotation strategy.
October 30, 2025 at 3:39 PM
francoissabot.bsky.social
No comment...
October 14, 2025 at 1:37 PM
francoissabot.bsky.social
A l'inauguration du Cluster PostGenAI de SCAI @sorbonne-universite.fr pour l' @ird-fr.bsky.social . L'IA est à la fois une chance pour le développement mais aussi un objet de recherche, à étudier et à utiliser (dans la mesure de l'éthique sociale et environnementale) #AI
Salle des Cordeliers à Paris, avec des fenêtres en ogive
April 1, 2025 at 4:13 PM
francoissabot.bsky.social
And it is much faster than the state-of-the-art tool for linear transfer LiftOff

4/5
Transfer rate comparison between liftoff and grannot on 1 to 12 genome of rices. Grannot is much faster
March 3, 2025 at 9:51 AM
francoissabot.bsky.social
GrAnnoT is fast, relies on the graph structure itself, and ensures a biologically relevant transfer of annotation: the frameshift rate induced by the annotation transfer is significantly lower in CDS

3/5
the frameshift rate induced by the annotation transfer is significantly lower in CDS
March 3, 2025 at 9:51 AM
francoissabot.bsky.social
to the graph itself or to any of the embedded genome within!

2/5
The picture despites the different steps in GrAnnot: step 1: transfer of the annotation from the source genome to the graph. Step 2 , the transfer from the graph to any target genome
March 3, 2025 at 9:51 AM
francoissabot.bsky.social
@murielgrosb.bsky.social is presenting her HDR docentship right now ! @ird-fr.bsky.social #palms #datepalm
February 7, 2025 at 1:10 PM
francoissabot.bsky.social
I am attendant the Software Heritage 2025 Symposium at @unesconow.bsky.social for #opensoftware #swh2025
Unesco amphitheatrum People are Washington for beginning
January 29, 2025 at 1:11 PM
francoissabot.bsky.social
According to Dall-E, here is my office: #bioinformatics #pangenomics #ttrpg #cats
December 18, 2024 at 3:19 PM
francoissabot.bsky.social
Formation sur les logiciels libres #opensource aux services de valorisation de l' @ird-fr.bsky.social ce matin!
Training on #OpenSoftware for valorisation services this morning
December 18, 2024 at 8:44 AM