Moving beyond MARCO.

The use of imaging systems in protein crystallisation means that the experimental setups no longer require manual inspection to determine the outcome of the trials. However, it leads to the problem of how best to find images which contain useful information about the crystallisation experiments. The adoption of a deeplearning approach in 2018 enabled a four-class machine classification system of the images to exceed human accuracy for the first time. Underpinning this was the creation of a labelled training set which came from a consortium of several different laboratories. The MARCO classification model does not have the same accuracy on local data as it does on images from the original test set; this can be somewhat mitigated by retraining the ML model and including local images. We have characterized the image data used in the original MARCO model, and performed extensive experiments to identify training settings most likely to enhance the local performance of a MARCO-dataset based ML classification model.

Yerba mate (Ilex paraguariensis, A. St.-Hil.) de novo transcriptome assembly based on tissue specific genomic expression profiles.

BACKGROUND: The most common infusion in southern Latin-American countries is prepared with dried leaves of Ilex paraguariensis A. St.-Hil., an aboriginal ancestral beverage known for its high polyphenols concentration currently consumed in > 90% of homes in Argentina, in Paraguay and Uruguay. The economy of entire provinces heavily relies on the production, collection and manufacture of Ilex paraguariensis, the fifth plant species with highest antioxidant activity. Polyphenols are associated to relevant health benefits including strong antioxidant properties. Despite its regional relevance and potential biotechnological applications, little is known about functional genomics and genetics underlying phenotypic variation of relevant traits. By generating tissue specific transcriptomic profiles, we aimed to comprehensively annotate genes in the Ilex paraguariensis phenylpropanoid pathway and to evaluate differential expression profiles. RESULTS: In this study we generated a reliable transcriptome assembly based on a collection of 15 RNA-Seq libraries from different tissues of Ilex paraguariensis. A total of 554 million RNA-Seq reads were assembled into 193,897 transcripts, where 24,612 annotated full-length transcripts had complete ORF. We assessed the transcriptome assembly quality, completeness and accuracy using BUSCO and TransRate; consistency was also evaluated by experimentally validating 11 predicted genes by PCR and sequencing. Functional annotation against KEGG Pathway database identified 1395 unigenes involved in biosynthesis of secondary metabolites, 531 annotated transcripts corresponded to the phenylpropanoid pathway. The top 30 differentially expressed genes among tissue revealed genes involved in photosynthesis and stress response. These significant differences were then validated by qRT-PCR. CONCLUSIONS: Our study is the first to provide data from whole genome gene expression profiles in different Ilex paraguariensis tissues, experimentally validating in-silico predicted genes key to the phenylpropanoid (antioxidant) pathway. Our results provide essential genomic data of potential use in breeding programs for polyphenol content. Further studies are necessary to assess if the observed expression variation in the phenylpropanoid pathway annotated genes is related to variations in leaves' polyphenol content at the population scale. These results set the current reference for Ilex paraguariensis genomic studies and provide a substantial contribution to research and biotechnological applications of phenylpropanoid secondary metabolites.

Classification of crystallization outcomes using deep convolutional neural networks.

The Machine Recognition of Crystallization Outcomes (MARCO) initiative has assembled roughly half a million annotated images of macromolecular crystallization experiments from various sources and setups. Here, state-of-the-art machine learning algorithms are trained and tested on different parts of this data set. We find that more than 94% of the test images can be correctly labeled, irrespective of their experimental origin. Because crystal recognition is key to high-density screening and the systematic analysis of crystallization experiments, this approach opens the door to both industrial and fundamental research applications.