Chao Deng

A Little bit of myself

My name is Chao Deng. I am from China. Currently I study at University of Southern Callifornia.

I am a PhD candidate in Computational Biology and Bioinformatics at USC. During my PhD studies, I obtained a MS in Computer Science at USC. Before came to US, I finished my MS in Mathematics at Peking University in China. I finished my BS in Mathematics at Shandong University in China.

My research is about estimating the expected number of unseen species in a population, given partial observations. For example, Shakespeare's known work contain 884,647 words, which have 31,534 word types. If another amount of work by Shakespeare are discovered, say 884,647t words, how many new word types do we expect to observe? When t is equal to 1, the answer is supurisingly consistent. However, in the era of big data, emerging applications require large t, which present new challenges.

Useful links

Publications and codes

My publications on Google Scholar

Codes under development on Github


We developed two softwares preseq & preseqR to see the unseen

preseqR, an R package to predict the number of species represented at least r times in a random sample

preseq, a C++ package to predict the complexity of a genomic sequencing library


Smithlab, an awesome lab with a bunch of awesome people

Manchester united, one of the best soccer teams in the world

Dota2, the greatest game I ever played

rMarkdown, a convenient tool to put together two great things, R and Markdown. Here is an excellent tutorial by Prof. Peter Ralph.



Numbe of heros


Number of items


Online players


Prize pool for TI6

This art is from Anuxinamoon on Deviantart

I like to play video games in my spare time. Among all games I ever played, Dota2 has to be the best. For me, two key factors in a game are diversity and balance. Dota2 is an excellent example in archiving both game diversity and game balance. Below are numbers showing how diverse this game is: 10 online players are required to play the game; 113 heros are available to pick; 148 items can be equipped to enhance attributes of a hero. For such a complicated game system, there is no doubt that many people complain about the steep learning curve.

Meanwhile, Dota2 is surprisingly balanced once you get used to it. In the recent competition event TI6, more than 100 heros were selected in at least one game, and almost every item is used. Of course the balance can be not achieved by one step. Here are patches of Dota2 in history.

One thing surprising me this summer is the Dota 2 AI bot. Without much hard coding, The bot beat a pro player on 1v1 in TI. Considering the genuine complexity of the game system, I was shocked by the progress made in AI. This also intrigues my interests in the AI field. Currently, I am developing a deep reinforcement learning approach to train my own Dota 2 AI bot. Hopefully one day it can beat me as well.