Search

Abstract Background The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-of-the-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Abstract Background Accumulating evidence suggests that the human microbiome impacts individual and public health. City subway systems are human-dense environments, where passengers often exchange microbes. The MetaSUB project participants collected samples from subway surfaces in different cities and performed metagenomic sequencing. Previous studies focused on taxonomic composition of these microbiomes and no explicit functional analysis had been done till now. Results As a part of the 2018 CAMDA challenge, we functionally profiled the available ~ 400 subway metagenomes and built predictor for city origin. In cross-validation, our model reached 81% accuracy when only the top-ranked city assignment was considered and 95% accuracy if the second city was taken into account as well. Notably, this performance was only achievable if the similarity of distribution of cities in the training and testing sets was similar. To assure that our methods are applicable without such biased assumptions we balanced our training data to account for all represented cities equally well. After balancing, the performance of our method was slightly lower (76/94%, respectively, for one or two top ranked cities), but still consistently high. Here we attained an added benefit of independence of training set city representation. In testing, our unbalanced model thus reached (an over-estimated) performance of 90/97%, while our balanced model was at a more reliable 63/90% accuracy. While, by definition of our model, we were not able to predict the microbiome origins previously unseen, our balanced model correctly judged them to be NOT-from-training-cities over 80% of the time. Our function-based outlook on microbiomes also allowed us to note similarities between both regionally close and far-away cities. Curiously, we identified the depletion in mycobacterial functions as a signature of cities in New Zealand, while photosynthesis related functions fingerprinted New York, Porto and Tokyo. Conclusions We demonstrated the power of our high-speed function annotation method, mi-faser, by analysing ~ 400 shotgun metagenomes in 2 days, with the results recapitulating functional signals of different city subway microbiomes. We also showed the importance of balanced data in avoiding over-estimated performance. Our results revealed similarities between both geographically close (Ofa and Ilorin) and distant (Boston and Porto, Lisbon and New York) city subway microbiomes. The photosynthesis related functional signatures of NYC were previously unseen in taxonomy studies, highlighting the strength of functional analysis.

Abstract Background After years of concentrated research efforts, the exact cause of Crohn’s disease (CD) remains unknown. Its accurate diagnosis, however, helps in management and preventing the onset of disease. Genome-wide association studies have identified 241 CD loci, but these carry small log odds ratios and are thus diagnostically uninformative. Methods Here, we describe a machine learning method—AVA,Dx (Analysis of Variation for Association with Disease)—that uses exonic variants from whole exome or genome sequencing data to extract CD signal and predict CD status. Using the person-specific coding variation in genes from a panel of only 111 individuals, we built disease-prediction models informative of previously undiscovered disease genes. By additionally accounting for batch effects, we were able to accurately predict CD status for thousands of previously unseen individuals from other panels. Results AVA,Dx highlighted known CD genes including NOD2 and new potential CD genes. AVA,Dx identified 16% (at strict cutoff) of CD patients at 99% precision and 58% of the patients (at default cutoff) with 82% precision in over 3000 individuals from separately sequenced panels. Conclusions Larger training panels and additional features, including other types of genetic variants and environmental factors, e.g., human-associated microbiota, may improve model performance. However, the results presented here already position AVA,Dx as both an effective method for revealing pathogenesis pathways and as a CD risk analysis tool, which can improve clinical diagnostic time and accuracy. Links to the AVA,Dx Docker image and the BitBucket source code are at https://bromberglab.org/project/avadx/.