AnnoFly: annotating Drosophila embryonic images based on an attention-enhanced RNN model

Motivation In the post-genomic era, image-based transcriptomics have received huge attention, because the visualization of gene expression distribution is able to reveal spatial and temporal expression pattern, which is significantly important for understanding biological mechanisms. The Berkeley Drosophila Genome Project has collected a large-scale spatial gene expression database for studying Drosophila embryogenesis. Given the expression images, how to annotate them for the study of Drosophila embryonic development is the next urgent task. In order to speed up the labor-intensive labeling work, automatic tools are highly desired. However, conventional image annotation tools are not applicable here, because the labeling is at the gene-level rather than the image-level, where each gene is represented by a bag of multiple related images, showing a multi-instance phenomenon, and the image quality varies by image orientations and experiment batches. Moreover, different local regions of an image correspond to different CV annotation terms, i.e. an image has multiple labels. Designing an accurate annotation tool in such a multi-instance multi-label scenario is a very challenging task. Results To address these challenges, we develop a new annotator for the fruit fly embryonic images, called AnnoFly. Driven by an attention-enhanced RNN model, it can weight images of different qualities, so as to focus on the most informative image patterns. We assess the new model on three standard datasets. The experimental results reveal that the attention-based model provides a transparent approach for identifying the important images for labeling, and it substantially enhances the accuracy compared with the existing annotation methods, including both single-instance and multi-instance learning methods. Availability and implementation http://www.csbio.sjtu.edu.cn/bioinf/annofly/ Supplementary information Supplementary data are available at Bioinformatics online.

Construction of a cDNA-based microarray for Drosophila melanogaster: a comparison of gene transcription profiles from SL2 and Kc167 cells

We have constructed a DNA microarray that represents approximately 6900 of the estimated 13 598 genes in the Drosophila melanogaster genome. The microarray contains 5756 target cDNAs from the Berkeley Drosophila Genome Project, 1078 cDNAs from the National Institutes of Health Drosophila testis cDNA library, and 546 gene fragments that were amplified from genomic DNA. The methods for DNA amplification and microarray manufacture are presented. Academic researchers can obtain the microarray from the Canadian Drosophila Microarray Centre. To evaluate the utility of these arrays, we compared the gene transcription profiles of two commonly used Drosophila cell lines. Analysis revealed that 5412 spot pairs gave signals consistently above the average background in Kc167 cells, whereas 5636 spot pairs met this criterion in SL2 cells. When the expression profiles of the cell lines were compared, 1437 genes displayed at least a 1.5-fold difference, and 170 genes had a threefold or greater difference between the two cell lines. In each case, with respect to Kc167 when compared with SL2 cells, the number of genes that were upregulated was nearly equal to the number of downregulated genes. This result demonstrates that despite the similar embryonic derivation of both cell lines, their transcriptional profiles are very different.Key words: DNA microarray, Drosophila, transcriptional regulation, SL2, Kc167.

Ras1 Interacts With Multiple New Signaling and Cytoskeletal Loci in Drosophila Eggshell Patterning and Morphogenesis

Little is known about the genes that interact with Ras signaling pathways to regulate morphogenesis. The synthesis of dorsal eggshell structures in Drosophila melanogaster requires multiple rounds of Ras signaling followed by dramatic epithelial sheet movements. We took advantage of this process to identify genes that link patterning and morphogenesis; we screened lethal mutations on the second chromosome for those that could enhance a weak Ras1 eggshell phenotype. Of 1618 lethal P-element mutations tested, 13 showed significant enhancement, resulting in forked and fused dorsal appendages. Our genetic and molecular analyses together with information from the Berkeley Drosophila Genome Project reveal that 11 of these lines carry mutations in previously characterized genes. Three mutations disrupt the known Ras1 cell signaling components Star, Egfr, and Blistered, while one mutation disrupts Sec61β, implicated in ligand secretion. Seven lines represent cell signaling and cytoskeletal components that are new to the Ras1 pathway; these are Chickadee (Profilin), Tec29, Dreadlocks, POSH, Peanut, Smt3, and MESK2, a suppressor of dominant-negative Ksr. A twelfth insertion disrupts two genes, Nrk, a “neurospecific” receptor tyrosine kinase, and Tpp, which encodes a neuropeptidase. These results suggest that Ras1 signaling during oogenesis involves novel components that may be intimately associated with additional signaling processes and with the reorganization of the cytoskeleton. To determine whether these Ras1 Enhancers function upstream or downstream of the Egf receptor, four mutations were tested for their ability to suppress an activated Egfr construct (λtop) expressed in oogenesis exclusively in the follicle cells. Mutations in Star and l(2)43Bb had no significant effect upon the λtop eggshell defect whereas smt3 and dock alleles significantly suppressed the λtop phenotype.

The Berkeley Drosophila Genome Project Gene Disruption Project: Single P-Element Insertions Mutating 25% of Vital Drosophila Genes

A fundamental goal of genetics and functional genomics is to identify and mutate every gene in model organisms such as Drosophila melanogaster. The Berkeley Drosophila Genome Project (BDGP) gene disruption project generates single P-element insertion strains that each mutate unique genomic open reading frames. Such strains strongly facilitate further genetic and molecular studies of the disrupted loci, but it has remained unclear if P elements can be used to mutate all Drosophila genes. We now report that the primary collection has grown to contain 1045 strains that disrupt more than 25% of the estimated 3600 Drosophila genes that are essential for adult viability. Of these P insertions, 67% have been verified by genetic tests to cause the associated recessive mutant phenotypes, and the validity of most of the remaining lines is predicted on statistical grounds. Sequences flanking >920 insertions have been determined to exactly position them in the genome and to identify 376 potentially affected transcripts from collections of EST sequences. Strains in the BDGP collection are available from the Bloomington Stock Center and have already assisted the research community in characterizing >250 Drosophila genes. The likely identity of 131 additional genes in the collection is reported here. Our results show that Drosophila genes have a wide range of sensitivity to inactivation by P elements, and provide a rationale for greatly expanding the BDGP primary collection based entirely on insertion site sequencing. We predict that this approach can bring >85% of all Drosophila open reading frames under experimental control.

Lethal P-lacZ insertion lines expressed during pattern respecification in the imaginal discs of Drosophila

The imaginal discs of Drosophila are a useful experimental system in which we can study the origin and genetic determination of spatial patterns in development. This involves the separation of the disc-cell population into distinct lineage compartments, based on clonally transmitted expression states of a number of known selector genes. However, these commitments can be abrogated and the compartment boundaries redeployed, when repatterning occurs in cultured disc fragments. This has so far only been explained using the idea of positional information. The genetic basis of this property of the imaginal disc system and its relationship to compartments have not been identified. Here we have screened over 470 recessive lethal P-lacZ enhancer-trap insertions from the Berkeley Drosophila Genome Project for expression after cell death, which initiates pattern respecification in the imaginal discs. The positive lines obtained identify essential genes that may be important for pattern formation. Most show patterned imaginal disc expression, and many have maternal or zygotic effects on embryonic development. One is an allele of schnurri, a gene that encodes a component of the decapentaplegic (dpp) signal transduction pathway used for positional signalling in the embryo and in imaginal discs.

Zygotic Lethal Mutations With Maternal Effect Phenotypes in Drosophila melanogaster. II. Loci on the Second and Third Chromosomes Identified by P-Element-Induced Mutations

Screens for zygotic lethal mutations that are associated with specific maternal effect lethal phenotypes have only been conducted for the X chromosome. To identify loci on the autosomes, which represent four-fifths of the Drosophila genome, we have used the autosomal “FLP-DFS” technique to screen a collection of 496 P element-induced mutations established by the Berkeley Drosophila Genome Project. We have identified 64 new loci whose gene products are required for proper egg formation or normal embryonic development.