Sanjeev Narayan Sharma

Tuning of FIR filter transition bandwidth using fractional Fourier transform

The transition bandwidth of window-based FIR filters is proportional to the window main-lobe width, which in turn is proportional to the length of the window function. As such, transition bandwidth of FIR filters can be directly tuned by varying window length for on-line tuning applications. However, analysis of window functions in fractional Fourier domain, a generalization of Fourier domain, also establishes the dependence of window main-lobe width on the order of fractional Fourier transform (FRFT). Thus, an alternative methodology to tune the transition bandwidth, based on FRFT, is developed in this work. The proposed methodology is useful for frequency domain filtering and introduces a comparative ease in tuning by eliminating the need to re-compute the impulse response coefficients. Also, significant computational saving has been achieved using FRFT. However, it is observed that the direct approach can introduce a lot more adjustability in the transition bandwidth than the FRFT approach. Apart from Kaiser window, considered to be optimum for FIR filter design, another window with a high side-lobe fall-off-rate (SLFOR), viz, Parzen-cos6(πt) (PC6), has also been used in the proposed on-line filter tuning. Better performance of windows with high SLFOR in on-line sharpening is illustrated with the aid of simulation results.

An Adaptive Window Length Strategy for Eukaryotic CDS Prediction

Signal processing-based algorithms for identification of coding sequences (CDS) in eukaryotes are non-data driven and exploit the presence of three-base periodicity in these regions for their detection. Three-base periodicity is commonly detected using short time Fourier transform (STFT) that uses a window function of fixed length. As the length of the protein coding and noncoding regions varies widely, the identification accuracy of STFT-based algorithms is poor. In this paper, a novel signal processing-based algorithm is developed by enabling the window length adaptation in STFT of DNA sequences for improving the identification of three-base periodicity. The length of the window function has been made adaptive in coding regions to maximize the magnitude of period-3 measure, whereas in the noncoding regions, the window length is tailored to minimize this measure. Simulation results on bench mark data sets demonstrate the advantage of this algorithm when compared with other non-data-driven methods for CDS prediction.

Improved exon prediction with transforms by de-noising period-3 measure

Gene finding techniques in eukaryotic cells can be divided into two categories, viz. - model-dependent and model-independent. In model-independent category, transforms are commonly used to identify exons or genes present in DNA sequences. In this work, a Post-Processing Algorithm (PPA) for enhancing gene prediction features of transforms is developed. PPA compares the N/3 spectral components of DNA signal with the corresponding spectrum of period-3 suppressed DNA signal. In the N/3 spectrum of DNA sequences, the bases for which the difference between these two spectrums is within a predefined threshold level are marked as non-coding (introns) regions. In such regions the signal values are replaced by the difference signal of the two spectrums. This substitution suppresses the noise in the intronic regions of the N/3 spectrum; while the coding region (exonic) signals are not affected, resulting in de-noised period-3 measures. PPA has been applied to process the period-3 coefficients of Discrete Fourier Transform (DFT), Paired Spectral Content (PSC), and Modified Gabor Wavelet Transform (MGWT) methods to de-noise their period-3 measures. Performance of the algorithm has been evaluated on HMR195, Burset/Guigo570, and Asp67 datasets using Receiver Operating Characteristic (ROC) and specificity versus sensitivity curves. The PPA, while preserving the model-independent characteristic of transform based methods, improves the probability of correct prediction of the exonic regions.

Studies on Z-Window Based FIR Filters

As per classification of the window functions, the Z-windows are grouped in the category of steerable side-lobe dip (SSLD) windows. In this work, the application of these windows for the design of FIR filters with improved filter parameters has been explored. The numbers of dips with their respective positions in the side-lobe region have been compositely used to tailor the window shape. Filter design relationships have been established and included in this paper. Simultaneously, an application of these Z-window based FIR filters in designing two-channel quadrature mirror filter (QMF) bank has been presented. Better values of reconstruction and aliasing errors have been achieved in contrast to the Kaiser window based QMF bank.

A simple algorithm for gene prediction with improved noise suppression

A simple algorithm to improvise the identification of protein coding regions in DNA sequences by period-3 property is presented. Background noise present in the period-3 DNA spectrum has been captured using DFT and notch filter. Elimination of this noise from the DNA spectrum improvised the detection of protein coding regions. Algorithm is data independent as it does not requires the empirical determination of any parameter for increasing the discrimination between coding and non-coding regions of a DNA sequence. An improvement in the global accuracy over the DFT and recently reported boosting approach is observed.

Identification of Microsatellites in DNA Using Adaptive S-Transform

Microsatellites are tandem repeats of size 1-6 base pairs, associated with various diseases, DNA fingerprinting, and also useful in evolutionary studies. A signal processing algorithm for microsatellite detection, based on adaptive S-transform is proposed. The standard deviation of the Gaussian window kernel of the S-transform has been optimized for integer periods of interest by maximizing the concentration measure. The time-frequency plot is generated using optimal standard deviation values. Candidate repeats are marked by comparing the spectrogram values in the time-frequency plot with a threshold. A preprocessing phase followed by a verification phase extracts final results from the candidate repeats. Simulation studies on DNA sequences establish the superiority of this algorithm over other existing methods. Applicability of this algorithm in the analysis of DNA sequences associated with repeat expansion diseases has also been demonstrated.

Design of FIR filters with better performance using Z-window

In this paper, a method for designing of FIR filters with very low pass band ripples (PBR) using Z-window has been proposed. The proposed filter design method is the derivative of Kaiser Window (KW). This method produces FIR filters featuring very low PBR and better stop band attenuation (SBA) with exactly marked passband and stopband frequencies with tolerable higher filter order.

Tandem repeats detection in DNA sequences using Kaiser window based adaptive S-transform

Abstract In computational biology the development of algorithms for the identification of tandem repeats in DNA sequences is a challenging problem. Tandem repeats identification is helpful in gene annotation, forensics, and the study of human evolution. In this work a signal processing algorithm based on adaptive S-transform, with Kaiser window, has been proposed for the exact and approximate tandem repeats detection. Usage of Kaiser window helped in identifying short as well as long tandem repeats. Thus, the limitation of earlier S-transform based algorithm that identified only microsatellites has been alleviated by this more versatile algorithm. The superiority of this algorithm has been established by comparative simulation studies with other reported methods.