Monday, September 05, 2016

FlowPy: from code to software


Cheers!! FlowPy, the software that we developed, got  three citations in published papers (PMID: 26116575, PMID: 27253695 & 27084942). 

It started as a lazy B. Tech project. My Ph. D student was using flow cytometry. We had good data. We wanted to use some advanced statistical analysis of that. Unfortunately, those tools are only available in commercial flow cytometry software. And to my utter surprise, our flow cytometer does not allow us to export raw data as plain-text. So we were trapped. We can not use any advanced statistical tools on our data.

So Tejas was entrusted to develop a software that would extract data from flow Cytometery files and export that as plain-text for further analysis. It was his B. Tech project. He wrote it in Python. We named it FlowPy.



At that time, I had no idea about software development. Software development is quite different from writing some codes for personal academic use. I had no idea about web-hosting of software, github, version control, GNU licence. We had no funding for the work either.

We were just happy to write the code, use it and share with others. So I posted the stuff at WikiDot, a blogging site. It is still there: http://flowpy.wikidot.com/

Over the years, other B. Tech students joined the team. Two students, M. V. S. R. Sastry, and Revanth Sai Kumar made most contributions. From a mere Python script, FlowPy graduated to a GUI-based software. It is no more a mere data extraction tool. It can also perform various statistical analysis and visualizations. 



We never published any paper on FlowPy. Just let it sleep at its WikiDot site. The information sipped slowly through the Web. Several Web lists of Flow Cytometry software listed FlowPy. Occasionally, I got queries on bug in our code. 

Slowly we were losing interest in FlowPy. For last one year, we stopped further development of FlowPy. But then came the bravo moment!! I realized some of our colleagues across the globe are using it successfully and two papers cited it. Unfortunately, WikiDot does not provide file download data. But Google Analytics tells me that we do get regular traffic. That too from all over the globe. 



Publishing a software is quite different from publishing a paper. A software is not a mere collection of information. It is a product: a product that has to be stable, reliable and works as promised. It has to work every time a user uses it. And we don't know the users.

That's why we are thrilled to see that some of our fellow scientists are finding FlowPy useful for their work. FlowPy is still buggy. It still has ample rooms for improvements.

We have again started development of FlowPy. Hope to remove the bugs and increase number of  in-built tools. If you use Flow Cytometry, give FlowPy a try and let us know your experience and expectations.

Happy Flow!

(Updated on 30/9/2016)

Sites that list FlowPy:

1. https://pypi.python.org/pypi/FlowCytometryTools/0.4.5
2. https://www.cds.caltech.edu/~murray/wiki/index.php/Flow_cytometry_software
3. http://gorelab.bitbucket.org/flowcytometrytools/
4. http://flowpy.software.informer.com/
5. http://en.freedownloadmanager.org/Windows-PC/FlowPy-FREE.html

Friday, August 26, 2016

Reject, minor, major revision and the fourth option

Once a reviewer of one of our papers wrote that she/he did not understand our mathematical model. That's a honest submission. No one expect that every one will understand everything. But had not that affected the decision made on our paper? May be. May be not.

But every reviewer faces this problem. As science is getting more and more interdisciplinary, one often find some part of the paper bit difficult to understand and review. I am not talking of complete ignorance. Neither talking of a badly written paper. Am talking of a situation where you broadly understand the concepts and issues, but lacks clarity on particulars in that paper. The best option then is to ask the authors to explain and help you understand their paper better.

Once you have understood the paper, with clarity, then only you can make a rational judgment on the paper. Isn't that obvious? But not in practise. Journals does not allow you to post queries or make comments on a manuscript without making a judgment out of three choices: minor revision, major revision, or reject. 

There is no scope of a dialogue, albeit with anonymity, between the people who did the science and those who did the vetting. Yes, there exist the practise of post-review rebuttal. But that's only after the reviewer has made the decision. 

The purpose of publishing scientific papers has changed with time. So has changed the practise and culture of peer-review. Journal editors complain of shortage in serious reviewers, authors complain of lackluster reviews,  reviewers complain of lack of professional incentive in reviewing papers.

Even then, there are people who review each others papers and do that with all earnest. They still believe in the elementary purpose of peer-review of a scientific paper: to improve the manuscript and to improve the work reported there.

Won't it be wiser to help this lot scientists to do the job better? One step towards better review would be to provide a fourth option to a reviewer. Let the reviewers post questions or start a thread of discussion with the authors, before they decide on the manuscript.

Obviously, such interactions would be considered as part of the review documents and has to bounded by a specific duration. It would also be bounded by all legal and ethical guidelines of peer-review.

Am not sure, how many of my peers will use this option. But letting some use it judiciously, wont harm science, but make it better.

Saturday, May 14, 2016

Everybody loves an anti-cancer drug

There are over 19000 papers, published till date, with the word anti-cancer in title or abstract. Over the years, funding for research on newer anti-cancer drugs has increased. So is the publications with this phrase (See the figure below). This phrase also have some magical power. It helps me to easily justify my research and grab a slice of funding pie. Unfortunately, the pie is never enough for all. 

Unfortunately, we are still far from wining the disease. 



Plot showing trend in publication of papers on "anti-cancer". Pubmed was searched for all the papers having the phrase "anti-cancer" either in abstract or in title. The numbers in parenthesis show the year of first report.

Working for drugs against cancer has some technical advantages over other diseases. Think about developing a new drug for an infectious disease, like Dengue or Malaria. It’s difficult to have a good in vitro model for many infectious diseases. When you have one, you need special laboratory facility and legal clearances to work on those. 

Cancer research has no such troubles; at least at the early stages of the project. Most of the in vitro assays are performed on cell lines. HeLa was the first human cell line, reported in 1952. Since then cell lines are the workhorse of anti-cancer drug development. These cells are treated with a drug and its ability to kill these cells is measured. Some time the drug does not kill the cell but just stops the cell division. That’s good enough for us. Measuring such cytotoxic or cytostatic effect of a drug is not so difficult. We have many cheap and reliable assays for this. One such is the famous MTT assay.

These experiments are simple, cheaper and you don’t have much legal and ethical issues. For a scientist, these are critical determinants. Social priority, science policy, academic fashion and ease of preliminary experiments, all these are behind the exponential growth in publications on anti-cancer agents.

But what is an anti-cancer agents? A search through the Pubmed throws up curious mix of items: plant extracts, nanomaterials to  atmospheric gas plasma. Most of these studies involve some form of in vitro cell culture-based experiment to show that these materials kill the cells, preferably through apoptosis. Essentially, the authors are checking cytotoxicity of these agents. Interestingly, some of these materials are also toxic to bacteria and often promoted as bactericidal agents, albeit in separate papers.

Most of these anticancer agents never makes to next step of evaluation. No body chase them further, not even the inventors. Authors move to another project, on another anti-cancer drug. Another paper is minted with the same key word. 

Cancer is a cellular disease. It is a disease with cells having genetic, epigenetic and phenotypic changes. To treat, either we have to convert these cells back to normal  or we have to get rid of them. For the time being, the first one seems improbable and our focus is on the other option.

In 1947, Sydney Farber used the same principle, when he used  aminopterin to treat children with leukemia. Aminopterin stops cancer by blocking cell division. Chemotherapeutic agents, developed subsequently, have the same property. They block proliferation of human cells through diverse mechanisms. Blockage of proliferation hits cancer cells and any other rapidly dividing normal cells. So, these drugs have some sort of inbuilt specificity: they block cell division and affect dividing cells more than those seating idle. 

However, many anti-cancer agents reported in academic literature do not follow the same logic. Something can be cytotoxic for different reasons. It may kill cells by blocking essential  processes like protein production. Cells can be killed by forming pores on the membrane or by oxidative damage. Many so called anti-cancer agents kill cells by these mechanisms. These methods have no specificity towards cancer cells and would affect every other cells in body. Even then, authors call those as anti-cancer agents. 

In fact. we really don’t have shortage of such non-specific cytotoxic or cytostatic agents. I will say, we have enough of such arsenals; enough to stop further search for new one. The focus should be more on developing strategies to deliver those specifically to cancer cells, sparing the normal one.

Academic research has its own dynamics. Some works on basic “blue sky” questions on how nature works. Others prefer to work on issues that has immediate social relevance. Discovery of a new cancer drug would have immediate social impact. Many of us may have such high goal, but we are mostly lost in closed alleys. 

Drug development is always an uncertain endeavor. Something that worked well in vitro may fail miserably in animal experiments or in clinical trials. Even then, our efforts should start with clear logic. Our strategy should have clear rationality based on our existing knowledge of other ant-cancer drugs. Unfortunately "logic" is loosing to the rush to get published. It is loosing to the fashion in academics. 

As the rogue cells keeps dividing within millions of people, we keep trying new methods to checkmate them. We keep trying, often, even without rationality. And the printing press churn out "Anti-cancer" in black and white.