Detecting DNA of novel fungal pathogens using ResNets and a curated fungi-hosts data collection

Abstract

Background Emerging pathogens are a growing threat, but large data collections and approaches for predicting the risk associated with novel agents are limited to bacteria and viruses. Pathogenic fungi, which also pose a constant threat to public health, remain understudied. Relevant data remain comparatively scarce and scattered among many different sources, hindering the development of sequencing-based detection workflows for novel fungal pathogens. No prediction method working for agents across all three groups is available, even though the cause of an infection is often difficult to identify from symptoms alone. Results We present a curated collection of fungal host range data, comprising records on human, animal and plant pathogens, as well as other plant-associated fungi, linked to publicly available genomes. We show that it can be used to predict the pathogenic potential of novel fungal species directly from DNA sequences with either sequence homology or deep learning. We develop learned, numerical representations of the collected genomes and visualize the landscape of fungal pathogenicity. Finally, we train multi-class models predicting if next-generation sequencing reads originate from novel fungal, bacterial or viral threats. Conclusions The neural networks trained using our data collection enable accurate detection of novel fungal pathogens. A curated set of over 1400 genomes with host and pathogenicity metadata supports training of machine-learning models and sequence comparison, not limited to the pathogen detection task. Availability and implementation The data, models and code are hosted at https://zenodo.org/record/5846345, https://zenodo.org/record/5711877 and https://gitlab.com/dacs-hpi/deepac. Supplementary information Supplementary data are available at Bioinformatics online.

AI-based Dynamic web server for real-time classification of raw genome sequences of ESKAPEE pathogens

“…Figure 8 shows the screenshot image of develop dynamic web server. We found 2 most closely related works [18][19] in the direction of bacterial classification using WGS data similar to our work, hence we selected these for comparison with the proposed work.…”

“…With our work, we are reporting the performance of model over a huge test data of more than 12,000 whole genome sequences, whereas models trained over more than 8000 whole genome sequences, i.e., total size of the data is more than 20,000 SRA whole genome sequences. Moreover, we have made our model up on a dynamic web server so that it can be tested or challenged with respect to its performance quoted anytime by anyone, which is missing in these works [18][19]. Our web server is blind-tested over any type of sequences by a few volunteers, and we reported that performance too.…”

“…It scores better being trained on SRA data generated by a variety of sequencers including Illumina and Nanopore sequencers as considered by [18]. The work in [19] is about identifying bacteria, virus, or fungi from an input whole genome sequence using deep learning (accuracy<95%) whereas with our work which not only identifies fungi (accuracy>98%), but also indicates which 7 types of ESKAPEE bacteria it is with accuracy>94.9%. The size of dataset in this work is 332 positive and 622 negative samples, which were further divided into train/test/validation sets, a deep learning model evolved over such a small dataset may lack generalization.…”

“…We propose to augment our system as the next step of a sequencer generating raw sequences to classify them into ESKAPEE pathogen in real-time as a modern diagnostic solution. As we trained our models over SRA data produced by a variety of sequencing devices, it is not dependent or biased for a specific sequencer as reported in earlier works [18][19]. The major upheaval from the existing systems is that it takes the bacterial classification process from wet lab to entirely in-silico labs.…”

AI-based Dynamic web server for real-time classification of raw genome sequences of ESKAPEE pathogens

“…We propose to augment our system as the next step of a sequencer generating raw sequences to classify them into ESKAPEE pathogens in realtime as a modern diagnostic solution. As we trained our models over SRA data produced by a variety of sequencing devices, it is not dependent or biased for a specific sequencer as reported in earlier works [18][19]. The major upheaval from the existing systems is that it takes the bacterial classification process from wet lab to entirely in-silico labs.…”