Computational Genomics with Privacy, Preserving Similar Patient Query

[*] Install CD-HIT

---

This project spins up a local server on the machine to simulate SPQ from client (doctor) to server (database/hospital)

---

Compile

Compile the PSI project

This project uses PSI project directly.

cd ComputationalGenomics/SecureApproxEditDistance/PSI
make clean
make

Copy executable file to the folder SecureApproxEditDistance. Specifically, in the folder ComputationalGenomics/SecureApproxEditDistance, execute

cp PSI/*.exe ./

Add executable permission to all bash scripts. In the folder ComputationalGenomics and ComputationalGenomics/SecureApproxEditDistance, execute

chmod +x *.sh
chmod +x *.py

Compile the Java code

cd ComputationalGenomics/SecureApproxEditDistance
mvn clean
mvn compile

Parameter Tuning for Previous Works

Accuracy & Efficiency

In the folder ComputationalGenomics, execute

snakemake

The accuracy result is in ComputationalGenomics/test/*.acc, and efficiency result is in ComputationalGenomics/test/*.result

If you want to generate the *.acc and *.result on your own, delete them before you snakemake:

rm -rf test/*.result
rm -rf test/*.acc

Result file format

• Accuracy : each .acc file contains 20 lines, the line number corresponds to the value $k$. The file name specifiies the method ( shingles / kbanded ) used, and the parameters ( gram size / banded value ) of the protocol.

• Efficiency : each .result file contains the standard output when running the protocol, which contains the running time. The file name specifies the method ( shingles / kbanded ) used, and the k value for top-k query, and parameters ( gram size / banded value ) of the protocol.

Our Protocol: Hierarchical Clustering

The accuracy of our protocol is obtained by the following command.

• Shingle-based Protocol on 50-sequences Database

  In the folder ComputationalGenomics/SecureApproxEditDistance, execute

  ./shingles_acc.sh ../db_cluster/db_cluster query.fa ./MapClusterToLabel_db.py db.fa db
  

  The result is in result_db_{2,4,...16}.txt.

• Shingle-based Protocol on 500-sequences database

  In the folder ComputationalGenomics/SecureApproxEditDistance, execute

  ./shingles_acc.sh ../db1_cluster/db1_cluster query1.1 ./MapClusterToLabel.py db1.fa db1
  

  The result is in result_db1_{2,4,...16}.txt.

• Kbanded-based Protocol on 50-sequences database

  In the folder ComputationalGenomics/SecureApproxEditDistance, execute

  ./kbanded_acc.sh ../db_cluster/db_cluster query.fa ./MapClusterToLabel_db.py db.fa db
  

  The result is in result_k_banded_db_{1,3,5,7,9}.txt.

• Kbanded-based Protocol on 500-sequences database

  In the folder ComputationalGenomics/SecureApproxEditDistance, execute

  ./shingles_acc.sh ../db1_cluster/db1_cluster query1.1 ./MapClusterToLabel.py db1.fa db1
  

  The result is in result_k_banded_db1_{1,3,5,7,9}.txt.

The script will print the result to the standard output and the file stdout in the current folder.

---

[*] In the cdhit-master folder, use the makefile command to make the executables and execute the following commands on gradx cluster.

./cd-hit-est -i db.fa -o db_cluster.fa -T 12 -M 1024

./cd-hit-est -i db1.fa -o db1_cluster.fa -T 12 -M 1024 -c 0.875 -n 6

---

Further, our testing results can be found in the test/ folder of which the genParams.sh script will run testing.