an easy CASSPER interface

CASSPER is a deep learning tool for automating the time-consuming Particle Picking task in obtaining the 3D reconstruction of protein structures using CryoEM micrographs. This work was published in Nature Communications in 2020.

  The updated pre-trained version of the CASSPER protein particle picking algorithm with the detailed manual is available here. This new version makes it easy to use CASSPER in just two lines of code in python.


Computers have become part of everyday life. The internet has made it even more versatile. It has become somewhat impossible to live without mobile phones. Although often forgotten, it is the software that gives life to all these machines. Without software, the most powerful supercomputer is as good as the lifeless body of a celebrity. In the Indian mythology there is a concept called Trans Migration which allows one soul to move to another body and perform a task. At this time the host remains unaware of what is going on. When the soul leaves the body, the host wakes up without any knowledge of what it has been performing under the influence of the soul that had migrated into its body. May be this reminds us of computer viruses. The major difference is that computer viruses permanently damages the software in the host and stays active until it is removed. In trans migration however, there is no system damage and the system recovers as soon as the migrated soul leaves. So the interesting question is the following: Can we have a similar situation in computing? Usually it takes a long time to install all the necessary software and set up a system to use. What if an already knowledgeable software soul migrate into the system and perform all the task as if it is all locally set up, finish the job and leave the system intact? That is exactly what ATMA or the Automated Trans Migration Algorithm does. It was originally created by Ninan in 2007 when his students told him that they do not have computers at home. The college had computers but they were all preinstalled with software that none is allowed to manipulate. So, ATMA was created taking some features from the Slax version of Linux Operating system. It was called Slax-ATMA and was able to boot into any personal computer, work from memory and exit without leaving a trace in the host machine. Since USB pen drives were inexpensive, it gave students a personal computer with everything they needed to work with. Slax-ATMA even had HPC versions that could configure Beawoulf clusters on the fly. Slax-ATMA gave way to Porteus-ATMA and of late to Ubuntu and Manjaro flavors of Linux.It has almost everything one needs to do scientific computing on even a thin client or a windows machine without installing any software on it!

You can find more about atma-BUNTU here  and about ATMANJARO here.

ATMANJARO ISO file may be downloaded from here. Or, the script that can be used to create  atma-BUNTU  is here.
Automated Trans Migration Algorithm carries the soul of scientific computing on a USB drive.

DBNN – The Difference Boosting Neural Network

DBNN is a Bayesian Classifier that uses the concept of Imposed Conditional Independence to reduce the computational overhead typically associated with Bayesian formalism. It has been widely used in various astronomy data challenges like star-galaxy classification, Quasar candidate identification, Gravitational Wave detector anomaly and glitches detection etc. The code in C++ is opensource and can be downloaded from here

The GitHub page also share the data used for creating the Quasar catalog, the catalog itself and the procedures to do similar classification problems using DBNN. The GIF animation to the right shows the 3D cube of about 6 million objects that include Quasars, similarly appearing Galaxies and stars from the colour space of the Sloan Digital Sky Survey

The animations show the 3D colour cube of about 6 million objects in our Quasar catalogue. The axis are the u-g, g-r and r-i colours derived from the SDSS u,g,r,i filters.