Biology subjects

Shatabda, S.

Publications and source records attributed to Shatabda, S..

7 recordsLinked to original sources

bioRxiv2026

Learning Strand-Equivariant Tokenizations of DNA with MirrorDNA

DNA has two strands, and for most regulatory label prediction tasks a sequence and its reverse complement should get same answer. Yet DNA language models are tokenized, trained and tested on one strand at a time. We show that this makes them inconsistent. BPE and $k$-mer tokenizers split 92--100\% of transcription-factor motifs differently on the two strands, and NT-v2, DNABERT-2 and HyenaDNA change 25-31\% of their predictions when the test sequences are reverse-complemented. We introduce {\model}, a compact DNA language model with a learned tokenizer that guarantees mirror-symmetric token boundaries across reverse-complement sequences, regardless of its learned parameters. We prove that this symmetry holds through the whole model. As a result, the same model can give identical answers on both strands when the label does not depend on the strand, and can still tell the strands apart when it does, which no strand-blind model can learn. Counting only predictions that are correct on both strands, \model\ more than doubles the MCC of NT-v2, DNABERT-2 and HyenaDNA. With 2.3M parameters, it is more accurate than the Caduceus models trained on the same data, and not significantly different from NT-v2 (98M parameters) on Genomic Benchmarks or from DNABERT-2 (117M) on the Nucleotide Transformer tasks. It also stays accurate on inputs sixteen times longer than those it was trained on. The same model trained with reverse-complement augmentation instead still changes about one prediction in ten with the strand, unless its outputs are averaged over both strands. Without any annotation, the tokenizer learns to keep the splice-donor signal \texttt{AGGT} in a single token. Code: https://github.com/NIRJHOR-DATTA/MirrorDNA.

bioinformatics↗

bioRxiv2021

Figbird: A probabilistic method for filling gaps in genome assemblies

MotivationAdvances in sequencing technologies have led to sequencing of genomes of a multitude of organisms. However, draft genomes of many of these organisms contain a large number of gaps due to repeats in genomes, low sequencing coverage and limitations in sequencing technologies. Although there exist several tools for filling gaps, many of these do not utilize all information relevant to gap filling. ResultsHere, we present a probabilistic method for filling gaps in draft genome assemblies using second generation reads based on a generative model for sequencing that takes into account information on insert sizes and sequencing errors. Our method is based on the expectation-maximization (EM) algorithm unlike the graph based methods adopted in the literature. Experiments on real biological datasets show that this novel approach can fill up large portions of gaps with small number of errors and misassemblies compared to other state of the art gap filling tools. Availability and ImplementationThe method is implemented using C++ in a software named "Filling Gaps by Iterative Read Distribution (Figbird)", which is available at: https://github.com/SumitTarafder/Figbird. Contactatif@cse.buet.ac.bd Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗

bioRxiv2021

PanClassif: Improving pan cancer classification of single cell RNA-seq using machine learning

Cancer is one of the major causes of human death per year. In recent years, cancer identification and classification using machine learning have gained momentum due to the availability of high throughput sequencing data. Using RNA-seq, cancer research is blooming day by day and new insights of cancer and related treatments are coming into light. In this paper, we propose PanClassif, a method that requires a very few and effective genes to detect cancer from RNA-seq data and is able to provide performance gain in several wide range machine learning classifiers. We have taken 22 types of cancer samples from The Cancer Genome Atlas (TCGA) having 8287 cancer samples and 680 normal samples. Firstly, PanClassif uses k-Nearest Neighbor (k-NN) smoothing to smooth the samples to handle noise in the data. Then effective genes are selected by Anova based test. For balancing the train data, PanClassif applies an oversampling method, SMOTE. We have performed comprehensive experiments on the datasets using several classification algorithms. Experimental results shows that PanClassif out-perform existing state-of-the-art methods available and shows consistent performance for two single cell RNA-seq datasets taken from Gene Expression Omnibus (GEO). PanClassif improves performances of a wide variety of classifiers for both binary cancer prediction and multi-class cancer classification. PanClassif is available as a python package (https://pypi.org/project/panclassif/). All the source code and materials of PanClassif are available at https://github.com/Zwei-inc/panclassif.

bioinformatics↗

bioRxiv2020

ACP-MHCNN: An Accurate Multi-Headed Deep-Convolutional Neural Network to Predict Anticancer peptides

Although advancing the therapeutic alternatives for treating deadly cancers has gained much attention globally, still the primary methods such as chemotherapy have significant downsides and low specificity. Most recently, Anticancer peptides (ACPs) have emerged as a potential alternative to therapeutic alternatives with much fewer negative side-effects. However, the identification of ACPs through wet-lab experiments is expensive and time-consuming. Hence, computational methods have emerged as viable alternatives. During the past few years, several computational ACP identification techniques using hand-engineered features have been proposed to solve this problem. In this study, we propose a new multi headed deep convolutional neural network model called ACP-MHCNN, for extracting and combining discriminative features from different information sources in an interactive way. Our model extracts sequence, physicochemical, and evolutionary based features for ACP identification through simultaneous interaction with different numerical peptide representations while restraining parameter overhead. It is evident through rigorous experiments using cross-validation and independent-dataset that ACP-MHCNN outperforms other models for anticancer peptide identification by a substantial margin. ACP-MHCNN outperforms state-of-the-art model by 6.3%, 8.6%, 3.7%, 4.0%, and 0.20 in terms of accuracy, sensitivity, specificity, precision, and MCC respectively. ACP-MHCNN and its relevant codes and datasets are publicly available at: https://github.com/mrzResearchArena/Anticancer-Peptides-CNN.

bioinformatics↗

bioRxiv2020

SubFeat: Feature Subspacing Ensemble Classifier for Function Prediction of DNA, RNA and Protein Sequences

The information of a cell is primarily contained in Deoxyribonucleic Acid (DNA). There is a flow of information of DNA to protein sequences via Ribonucleic acids (RNA) through transcription and translation. These entities are vital for the genetic process. Recent developments in epigenetic also show the importance of the genetic material and knowledge of their attributes and functions. However, the growth in known attributes or functionalities of these entities are still in slow progression due to the time consuming and expensive in vitro experimental methods. In this paper, we have proposed an ensemble classification algorithm called SubFeat to predict the functionalities of biological entities from different types of datasets. Our model uses a feature subspace based novel ensemble method. It divides the feature space into sub-spaces which are then passed to learn individual classifier models and the ensemble is built on this base classifiers that uses a weighted majority voting mechanism. SubFeat tested on four datasets comprising two DNA, one RNA and one protein dataset and it outperformed all the existing single classifiers and as well as the ensemble classifiers. SubFeat is made availalbe as a Python-based tool. We have made the package SubFeat available online along with a user manual. It is freely accessible from here: https://github.com/fazlulhaquejony/SubFeat.

bioinformatics↗

bioRxiv2019

iPromoter-BnCNN: a Novel Branched CNN Based Predictor for Identifying and Classifying Sigma Promoters

MotivationPromoter is a short region of DNA which is responsible for initiating transcription of specific genes. Development of computational tools for automatic identification of promoters is in high demand. According to the difference of functions, promoters can be of different types. Promoters may have both intra and inter class variation and similarity in terms of consensus sequences. Accurate classification of various types of sigma promoters still remains a challenge. ResultsWe present iPromoter-BnCNN for identification and accurate classification of six types of promoters -{sigma} 24,{sigma} 28,{sigma} 32,{sigma} 38,{sigma} 54,{sigma} 70. It is a CNN based classifier which combines local features related to monomer nucleotide sequence, trimer nucleotide sequence, dimer structural properties and trimer structural properties through the use of parallel branching. We conducted experiments on a benchmark dataset and compared with six state-of-the-art tools to show our supremacy on 5-fold cross-validation. Moreover, we tested our classifier on an independent test dataset. AvailabilityOur proposed tool iPromoter-BnCNN web server is freely available at http://103.109.52.8/iPromoter-BnCNN. The runnable source code can be found here. Contactrafeed@cse.uiu.ac.bd Supplementary informationSupplementary data (benchmark dataset, independent test dataset, model files, structural property information, attention mechanism details and web server usage) are available at Bioinformatics. online.

genomics↗

bioRxiv2019

DeepDBP: Deep Neural Networks for Identification of DNA-binding Proteins

DNA-Binding proteins (DBP) are associated with many cellular level functions which includes but not limited to bodys defense mechanism and oxygen transportation. They bind DNAs and interact with them. In the past DBPs were identified using experimental lab based methods. However, in the recent years researchers are using supervised learning to identify DBPs solely from protein sequences. In this paper, we apply deep learning methods to identify DBPs. We have proposed two different deep learning based methods for identifying DBPs: DeepDBP-ANN and DeepDBP-CNN. DeepDBP-ANN uses a generated set of features trained on traditional neural network and DeepDBP-CNN uses a pre-learned embedding and Convolutional Neural Network. Both of our proposed methods were able to produce state-of-the-art results when tested on standard benchmark datasets.DeepDBP-ANN had a train accuracy of 99.02% and test accuracy of 82.80%.And DeepDBP-CNN though had train accuracy of 94.32%, it excelled at identifying test instances with 84.31% accuracy. All methods are available codes and methods are available for use at: https://github.com/antorkhan/DNABinding.

bioinformatics↗