Shouji: a fast and efficient pre-alignment filter for sequence alignment
read more
Citations
A Modern Primer on Processing in Memory.
GenASM: A High-Performance, Low-Power Approximate String Matching Acceleration Framework for Genome Sequence Analysis
Processing-in-memory: A workload-driven perspective
In-DRAM Bulk Bitwise Execution Engine.
NERO: A Near High-Bandwidth Memory Stencil Accelerator for Weather Prediction Modeling
References
A general method applicable to the search for similarities in the amino acid sequence of two proteins
Identification of common molecular subsequences.
Binary codes capable of correcting deletions, insertions, and reversals
Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM
Related Papers (5)
Frequently Asked Questions (18)
Q2. What have the authors stated for future works in "Shouji: a fast and efficient pre-alignment filter for sequence alignment" ?
Another potential target of their research is to explore the possibility of accelerating optimal alignment calculations for longer sequences ( few tens of thousands of characters ) ( Senol et al., 2018 ) using pre-alignment filtering.
Q3. What are the main components of the hardware accelerator?
Hardware accelerators include multi-core and single instruction multiple data (SIMD) capable central processing units (CPUs), graphics processing units (GPUs) and field-programmable gate arrays (FPGAs).
Q4. What is the way to compute the optimal alignment?
The backtracking step required for the optimal alignment computation involves unpredictable and irregular memory access patterns, which poses a difficult challenge for efficient hardware implementation.
Q5. How many LUTs are required for a single MAGNET filtering unit?
(i) The design for a single MAGNET filtering unit requires about 10.5 and 37.8% of the available LUTs for edit distance thresholds of 2 and 5, respectively.
Q6. What is the way to use the available resources in the FPGA?
To make the best use of the available resources in the FPGA chip, their algorithm utilizes the operations that are easily supported on an FPGA, such as bitwise operations, bit shifts and bit count.
Q7. What is the fundamental computational step in bioinformatics analyses?
One of the most fundamental computational steps in most bioinformatics analyses is the detection of the differences/similarities between two genomic sequences.
Q8. What is the way to compute a sequence pair?
The multi-core architecture of CPUs and GPUs provides the ability to compute alignments of many sequence pairs independently and concurrently (Georganas et al., 2015; Liu and Schmidt, 2015).
Q9. What is the workflow of the accelerator?
The workflow of the accelerator starts with transmitting the sequence pair to the FPGA through the fastest communicationmedium available on the FPGA board (i.e. PCIe).
Q10. What is the importance of enumerating all possible prefixes?
Enumerating all possible prefixes is necessary for tolerating edits that result from both sequencing errors (Fox et al., 2014) and genetic variations (McKernan et al., 2009).
Q11. How can the upper E diagonals be implemented?
The upper E diagonals can be implemented by gradually shifting the pattern (P) to the right-hand direction and then performing bitwise XOR with the text (T).
Q12. How many false accept rates does MAGNET show for high-edit datasets?
(iv) MAGNET shows up to 1577, 3550 and 25 552 lower false accept rates for high-edit datasets (3.5, 14.7 and 135 for lowedit datasets) compared to GateKeeper and SHD for sequence lengths of 100, 150 and 250 characters, respectively.
Q13. What is the effect of MAGNET on the execution time of the aligner?
The authors observe that if the execution time of the aligner is much larger than that of the pre-alignment filter (which is the case for Edlib, Parasail and GSWABE for E ¼ 5 characters), then MAGNET provides up to 1.3 more end-to-end speedup over Shouji.
Q14. How many false accept rates does MAGNET show for highedit datasets?
MAGNET also shows up to 205, 951 and 16 760 lower false accept rates for highedit datasets (2.7, 10 and 88 for low-edit datasets) over Shouji for sequence lengths of 100, 150 and 250 characters, respectively.
Q15. What is the way to find the common subsequences?
One way of finding these common subsequences is to use a brute-force approach, which examines all the streaks of diagonally consecutive zeros that start at the first column and selects the streak that has the largest number of zeros as the first common subsequences.
Q16. How does Shouji improve the accuracy of pre-alignment filtering?
Shouji improves the accuracy of pre-alignment filtering by up to two orders of magnitude compared to the best-performing existing pre-alignment filter, GateKeeper.
Q17. How does the Shouji algorithm find the number of zeros in each 4-bit vector?
It quickly finds the number of zeros in each 4-bit vector using a hardware look-up table that stores the 16 possible permutations of a 4-bit vector along with the number of zeros for each permutation.
Q18. What is the way to make the use of existing aligners?
Shouji offers the ability to make the best use of existing aligners without sacrificing any of their capabilities, as it does not modify or replace the alignment step.