Classification of protein sequences by means of irredundant patterns

Clustering of reads with alignment-free measures and quality values

Background

The data volume generated by Next-Generation Sequencing (NGS) technologies is growing at a pace that is now challenging the storage and data processing capacities of modern computer systems. In this context an important aspect is the reduction of data complexity by collapsing redundant reads in a single cluster to improve the run time, memory requirements, and quality of post-processing steps like assembly and error correction. Several alignment-free measures, based on k-mers counts, have been used to cluster reads.

Quality scores produced by NGS platforms are fundamental for various analysis of NGS data like reads mapping and error detection. Moreover future-generation sequencing platforms will produce long reads but with a large number of erroneous bases (up to 15 %).

Results

In this scenario it will be fundamental to exploit quality value information within the alignment-free framework. To the best of our knowledge this is the first study that incorporates quality value information and k-mers counts, in the context of alignment-free measures, for the comparison of reads data. Based on this principles, in this paper we present a family of alignment-free measures called D^q-type. A set of experiments on simulated and real reads data confirms that the new measures are superior to other classical alignment-free statistics, especially when erroneous reads are considered. Also results on de novo assembly and metagenomic reads classification show that the introduction of quality values improves over standard alignment-free measures. These statistics are implemented in a software called QCluster (http://www.dei.unipd.it/~ciompin/main/qcluster.html).

Assembly-free genome comparison based on next-generation sequencing reads and variable length patterns

Background

With the advent of Next-Generation Sequencing technologies (NGS), a large amount of short read data has been generated. If a reference genome is not available, the assembly of a template sequence is usually challenging because of repeats and the short length of reads. When NGS reads cannot be mapped onto a reference genome alignment-based methods are not applicable. However it is still possible to study the evolutionary relationship of unassembled genomes based on NGS data.

Results

We present a parameter-free alignment-free method, called Under2¯, based on variable-length patterns, for the direct comparison of sets of NGS reads. We define a similarity measure using variable-length patterns, as well as reverses and reverse-complements, along with their statistical and syntactical properties. We evaluate several alignment-free statistics on the comparison of NGS reads coming from simulated and real genomes. In almost all simulations our method Under2¯ outperforms all other statistics. The performance gain becomes more evident when real genomes are used.

Conclusion

The new alignment-free statistic is highly successful in discriminating related genomes based on NGS reads data. In almost all experiments, it outperforms traditional alignment-free statistics that are based on fixed length patterns.

Beyond Fixed-Resolution Alignment-Free Measures for Mammalian Enhancers Sequence Comparison

The cell-type diversity is to a large degree driven by transcription regulation, i.e., enhancers. It has been recently shown that in high-level eukaryotes enhancers rarely work alone, instead they collaborate by forming clusters of cis-regulatory modules (CRMs). Even if the binding of transcription factors is sequence-specific, the identification of functionally similar enhancers is very difficult. A similarity measure to detect related regulatory sequences is crucial to understand functional correlation between two enhancers. This will allow large-scale analyses, clustering and genome-wide classifications. In this paper we present Under2, a parameter-free alignment-free statistic based on variable-length words. As opposed to traditional alignment-free methods, which are based on fixed-length patterns or, in other words, tied to a fixed resolution, our statistic is built upon variable-length words, and thus multiple resolutions are allowed. This will capture the great variability of lengths of CRMs. We evaluate several alignment-free statistics on simulated data and real ChIP-seq sequences. The new statistic is highly successful in discriminating functionally related enhancers and, in almost all experiments, it outperforms fixed-resolution methods. Finally, experiments on mouse enhancers show that Under2 can separate enhancers active in different tissues.

AVAILABILITY

http://www.dei.unipd.it/~ciompin/main/UnderIICRMS.html.

Alignment-free phylogeny of whole genomes using underlying subwords

Background

With the progress of modern sequencing technologies a large number of complete genomes are now available. Traditionally the comparison of two related genomes is carried out by sequence alignment. There are cases where these techniques cannot be applied, for example if two genomes do not share the same set of genes, or if they are not alignable to each other due to low sequence similarity, rearrangements and inversions, or more specifically to their lengths when the organisms belong to different species. For these cases the comparison of complete genomes can be carried out only with ad hoc methods that are usually called alignment-free methods.

Methods

In this paper we propose a distance function based on subword compositions called Underlying Approach (UA). We prove that the matching statistics, a popular concept in the field of string algorithms able to capture the statistics of common words between two sequences, can be derived from a small set of “independent” subwords, namely the irredundant common subwords. We define a distance-like measure based on these subwords, such that each region of genomes contributes only once, thus avoiding to count shared subwords a multiple number of times. In a nutshell, this filter discards subwords occurring in regions covered by other more significant subwords.

Results

The Underlying Approach (UA) builds a scoring function based on this set of patterns, called underlying. We prove that this set is by construction linear in the size of input, without overlaps, and can be efficiently constructed. Results show the validity of our method in the reconstruction of phylogenetic trees, where the Underlying Approach outperforms the current state of the art methods. Moreover, we show that the accuracy of UA is achieved with a very small number of subwords, which in some cases carry meaningful biological information.

Availability

http://www.dei.unipd.it/∼ciompin/main/underlying.html

The irredundant class method for remote homology detection of protein sequences

The automatic classification of protein sequences into families is of great help for the functional prediction and annotation of new proteins. In this article, we present a method called Irredundant Class that address the remote homology detection problem. The best performing methods that solve this problem are string kernels, that compute a similarity function between pairs of proteins based on their subsequence composition. We provide evidence that almost all string kernels are based on patterns that are not independent, and therefore the associated similarity scores are obtained using a set of redundant features, overestimating the similarity between the proteins. To specifically address this issue, we introduce the class of irredundant common patterns. Loosely speaking, the set of irredundant common patterns is the smallest class of independent patterns that can describe all common patterns in a pair of sequences. We present a classification method based on the statistics of these patterns, named Irredundant Class. Results on benchmark data show that the Irredundant Class outperforms most of the string kernels previously proposed, and it achieves results as good as the current state-of-the-art method Local Alignment, but using the same pairwise information only once.