musings on music and life

October 31, 2016

Review of MITx 6.00.1x

Filed under: Coding, education — sankirnam @ 12:53 pm

just finished the above course – I just completed the last problem on the final exam and completed the exit survey a few minutes ago, so I figured I would write my thoughts on the course while they’re still fresh.

My impressions of the course are unanimously positive. I just finished the current iteration of the course (Aug – Nov 2016), and I found it to be excellent. I just finished writing an email to Prof. Grimson (the professor conducting the course), thanking him for all his efforts in preparing such high-quality materials!

Keep in mind that the title of the course is “Introduction to Computer Science and Programming using Python”, and so it is aimed to be an intro CS course of sorts. Nonetheless, it does serve as a very good introduction to the Python language, and covers fundamental CS concepts while teaching the Python language, including the various data structures (lists, tuples, and dictionaries), functions, and classes. The course isn’t intended to teach Python specifically, and so doesn’t cover a lot of the things unique to Python (such as lambda functions, list comprehensions, and other topics).

In retrospect, I wish I had taken this course before taking the “Data Science” bootcamp this summer – I would have been better prepared and would have had at least a rudimentary understanding of the CS fundamentals. Anyway, what’s done is done, and I’m glad that I was able to take this course.

The problem sets were very well crafted. They were appropriately challenging, and I probably did spend around the recommended 15 hours/week or so on them, and they weren’t too difficult where I would have ended up throwing my computer out the window and quitting in frustration. The bonus is that I ended up also learning how to use my computer better – since this course uses Python 3, I ended up using Anaconda to install that (so that I could manage that alongside my existing Python 2 install). I also ended up using Spyder as my IDE of choice for the course, and I’ve come to like that a lot.

As always, if you want to take a look at the problem sets, exercises, and my solutions, I’m posting everything to my github.

Proof of completion (I blacked out my username and email to dodge spambots):mitx6001x

Anyway, onwards to the sequel course, 6.00.2x! I started this course and it’s proving to be MUCH tougher, since the barrier is no longer the Python language, but abstractly developing algorithms before implementing them in Python.

October 18, 2016

Still on the job search

Filed under: Chemistry Jobs — sankirnam @ 4:09 pm

It has been about 20 months since I got my PhD in chemistry, and I’m still on the search applying for my first job. It’s been a really emotionally draining, tough ride. Before I graduated, I had heard horror stories from others about the chemistry job market and how brutal it is…but there’s nothing like experiencing it firsthand yourself. There are several major hurdles, which I’ll try and document here.

  • Applying online. The major portal for job applications is now online. This is convenient for both job seekers and employers; job seekers can electronically send applications for positions (which normally include a cover letter, resume and/or supplemental information such as a research summary) from the comfort of their home or office. With the internet, employers and recruiters also have a larger talent pool. The process is still time-consuming, however; I would estimate that it takes me on average about 45 minutes to fill out an online application; this includes filling out the information in the online forms (I always end up having to manually do this since the resume parsing never works), making edits to my resume to tailor it for the position, and writing a cover letter. I put in all this work, only to be greeted with:
  • The cone of silence. This is the most frustrating aspect of the job search. You’ll submit your application online, and usually within 1 minute receive an e-mail saying “Thank you for your application, it has been successfully received, and will be reviewed by our team”. This will be followed by….. silence. You won’t hear anything for weeks, or even months on end. I have a list of all the jobs I have applied to, and at least 85% of them have a note saying “Status: No reply”. I would follow up… if I knew who to follow up with! The internet is only so helpful in this regard, and its not always possible to find out who the particular recruiter or hiring manager is for a particular position.
    Case in point: I recently applied to 3 positions in Allergan in August, and still have not heard anything back. The recruiter for the position (as listed on LinkedIn) was unresponsive to my e-mails, and it was only by following up with a friend of a friend in the company that I was informed that yes, they had my application and that it was still under consideration. The funny thing is, these positions are still being listed on job boards and are still accepting applications!
  • The insane saturation of this particular job market. Don’t listen to the politicians – we don’t have a shortage of scientists in this country. We have a massive, massive, glut, and anyone who does any kind of scientific hiring will be able to corroborate this. It’s especially bad at the PhD level – back to my example at Allergan, I was discreetly informed not to get my hopes up since they received 500+ applications for 1 opening in medicinal chemistry. Plus, I did happen to have some nice chats with senior executives at [unnamed pharma companies], and they (somewhat condescendingly) told me to stop wasting my time, because pharma hiring is focused on pedigree; if you don’t have a degree from Harvard/Stanford/MIT/Caltech, etc. your application will be immediately discarded. The irony is that these executives did not have degrees from those schools.

That being said, it seems to me that there’s really only one surefire way to get a job out of school, and that is through campus recruiting. Unfortunately no companies in my area of study (chemistry) came to hire at my university, so that ruled out that approach. The other way is to join a position through a friend’s referral, which works for smaller companies and startups. Applying to big companies is seemingly slower, since the application has to go through several stages – a recruiter (who may or may not know the subject and understand your resume), followed by an interview with the hiring manager (who will be knowledgeable in the domain), and further interviews. I have been told that ‘80% of jobs never get advertised’ and other statistics like that, but those are only relevant for experienced job seekers looking to move laterally; it’s not relevant for fresh graduates looking for their first job. For your first job, you need to play by the company’s rules for applications. Once you get experience and make contacts, then you can get your friends to backdoor you into positions at other companies.

At least, that’s my observation. I don’t know what other avenues there are for gaining employment (I should specify that I mean relevant employment that would utilize my education and background; I could always go and be a cashier at a grocery store, but that would be a massive waste of my education and also the taxpayer money that went into funding that education). If anyone has any ideas, let me know!

The other question that comes up is “so, what about Data Science?”. Yes, that is still on the table; I’m still working with recruiters from Harnham, but nothing has panned out yet. To be honest, I’m not totally thrilled. After doing some soul-searching, I have realized that I don’t necessarily have the mentality to go into “Data Science” or software development. It’s one of those things that I am simultaneously overqualified (I have a PhD, after all) and underqualified (PhD in a irrelevant subject) for. I don’t have years of software development experience or deep knowledge of a lower-level language (e.g. Java, C, C++), or a degree in math, statistics, computer science, or physics (which are considered sexy in this field). I have actually been advised that maybe I should hide my PhD in organic chemistry since it is irrelevant to Data Science. At the same time, I’m glad I took that bootcamp – I can now study the material at my own pace, and the knowledge of programming and computational thinking is becoming increasingly critical today, what with the amount of time we spend interfacing with digital devices in the form of laptops, cellphones, tablets, and other computers.

So….I’m still looking for my first job in chemistry! If anyone has any leads, please do let me know.

I forgot to include this gem as an example to illustrate my point:
I applied to this position at BBraun in Irvine in March – on paper, it is a typical Analytical chemist position, and one that I am reasonably well-suited for. The only weird thing is that they explicitly want “Pharmaceutical industry or a relevant post-doc experience of 3-6 years for PhD”, which doesn’t make much sense (but can be chalked up to “credential inflation” in this over-saturated job market). In any case, I was swiftly rejected by the company, but to my surprise, the position is still up, over 6-7 months later! Stuff like this just really infuriates me. Companies like these waste so much time searching for the perfect “purple unicorn” candidate, and then raise a hue and cry about a “STEM shortage” when they’ve rejected everyone for the most random reasons.jackie-chan-wtf

I know people are curious, so here are the stats:

Jobs applied to: 1465
Interviews: 9

EDIT (10/26/2016): This morning, I was greeted with this e-mail from Merck: “Thank you for your interest in Merck.  We appreciate you taking the time to pursue career opportunities with us.  We have chosen at this time to suspend the search for this position and may reopen the search at a later date”. I applied to this position 2 months ago (August 25, 2016, to be exact), never heard anything back, and then received this notification. Seriously, something is screwy in hiring – has this happened to other people, or is it just me? Also, I honestly think there should be less of a stigma against unemployment – just look at how much time elapses in the job search! The companies are the ones that are slow in getting back to job seekers; in other words, the rate-determining step in the job search is waiting to hear back from companies, which means that individuals should not be held completely responsible for long periods of unemployment if they are applying aggressively.

2ND EDIT (11/16/2016): Yesterday, I got this email from Eli Lilly: “Thank you for your recent inquiry for the Research Scientist-Small Molecule Design and Development-Developability position, requisition #28370BR.

The position in which you originally expressed interest has unfortunately been cancelled and was not filled. Please feel free to review current openings and submit your interest accordingly”. At least this position didn’t leave me hanging for that long – I applied to it on 10/13/2016. I’m just completely nonplussed here…

October 13, 2016

NMR data in papers

Filed under: Chemistry — sankirnam @ 1:42 pm

As someone who has written several papers in organic chemistry, and is currently in the process of writing another one, I just had this thought:

Why can’t journals require authors to include the .fid files of NMR spectra in the supplemental information, as opposed to spectral data lists or printouts of the spectra?

Including printouts is a holdover from a bygone era; thanks to the internet, everything is now digital, and storage space is no longer an issue for most people (or companies). Journals should make authors include the .fid files of any NMR spectra required to accompany a publication! At best, they get up to 50-70 MB (for 2-D NMR spectra), which in this day and age is not that big.

Making authors include .fid files has several benefits:

  1. Writing papers becomes a lot less tedious. Yea, its not fun to sit at your computer and adjust the magnification, aesthetics, and other aspects of a spectrum when all you’re really interested in is the data. Yes, I know you want to make it look like a piece of art, but really, if you have pride in your skills as a scientist, and if you have any ethical integrity at all, you should be willing to stand behind the raw .fid files of any NMR spectra of your compounds.
  2. It’s actually easier for other people. Let’s face it – a lot of the times, the integral values and peak numbers are not very visible on a printout. Also, if you have a very complex spectrum (such as from a natural product), then including multiple zoomed-in regions or going into detail on every single multiplet is a hassle. Also, if someone is trying to reproduce your procedure, doing a .fid-to-.fid comparison is a lot easier than looking at printouts.
  3. It reduces the risk for fraud. The Bengü Sezen case could have been partly avoided if she had been forced to submit .fid files for the NMR data of her products, as opposed to doctored NMR spectra. It’s much, much harder to manipulate a raw .fid file than photoshop an NMR printout.

If only journals accepted .fid files – I could just put them all in a folder, upload it, and be on my way! But unfortunately, most journals do not. JOC (Journal of Organic Chemistry) still requires a list of spectral data for synthesized compounds in the experimental section of the manuscript as well as printouts of the spectra in the Supplemental Information. I guess manually entered spectral data is still required because NMR processing software is still not very good at identifying multiplets and picking peaks (it’s especially bad when they overlap). A trained eye will know what to look for, but a computer will not.

These are some of the challenges that lie in bringing organic chemistry to the 21st century!

October 12, 2016

Classics in Organic Chemistry, Part VII

Filed under: Classics in Organic Chemistry — sankirnam @ 6:04 pm

Ok, let’s keep the train rolling here…

This next topic is related to an ill-defined project I worked on early in my PhD, where I was investigating synthetic reactions related to isocyanide synthesis using TMSCN. One of the first places my advisor told me to look was at an intriguing 1982 JACS communication by Prof. Paul Gassman. To preface, it is well-established that the cyanide ion is ambident and can react from either the or the C position; the conditions employed can influence whether a nitrile or isonitrile will be obtained as the product. Gassman’s paper revisited this topic using epoxides as the electrophile, and he demonstrated that by intelligently choosing the right Lewis Acid (ZnI2 in this case), one can obtain β-isocyano alcohols as the product upon ring-opening with TMSCN (followed by desilylation with KF).cyanide1

Now that I reflect about the background for this paper, it is actually not as serendipitous as I had used to think as a first/second year grad student. Gassman had previously published a few papers, including an Organic Syntheses procedure, for converting ketones to cyanohydrins; the conditions employed in the above reaction are pretty much identical, save for switching out the ketone for an epoxide.

The utility of this reaction lies in the fact that isocyanides are extremely valuable synthons – there are a family of extremely useful multicomponent reactions based on isocyanides, including the Ugi Reaction and the Passerini Reaction. These reactions succeed because the isocyanide is a rare example of a (1,1)-amphoteric molecule; the same atom (the R-carbon) establishes a connection with both the nucleophile (carboxylic acid) and electrophile (aldehyde or imine). The Ugi reaction was developed by Prof. Ivar Ugi (no surprise), and I discovered a cool fact about him when I happened to check out his book Isonitrile Chemistry from the library, and I saw that it said “Prof. Ivar Ugi, University of Southern California”. Apparently he was a faculty member at USC for a short time (around a year or two) in the early 70’s, before Prof. Olah came to USC.

In any case, the next question is, how does this isocyanation work? In my mind, the mechanism is pretty straightforward; it’s simply a variant of the Ritter reaction with TMSCN, avoiding the use of aqueous acids to prevent hydration of the intermediate nitrilium ion or isocyanide to an amide. This reaction has seen a slow stream of contributions – you can see the references for a list of papers that describe the conversion of various types of compounds to isocyanides or amides. Recently, Ryan Shenvi (Scripps) revisited this chemistry and somehow got a paper in Nature; I don’t understand why this was selected for publication, because as you can see here, there’s nothing truly original about it, and the conditions are not really practical:

A solution of trifluoroacetate 13 (32.0mg, 0.1 mmol) in TMSCN(0.1ml) was cooled to 0 ℃ and treated with a solution of anhydrous Sc(OTf)3 (1.5 mg, 0.003 mmol) in TMSCN (0.1 ml). […]”

The reaction is carried out neatusing TMSCN as the solvent! Not really scalable, and only for the truly desperate.

If you ask me, the cyanation reactions are more intriguing, because the mechanism is more unclear. Prof. Weber (who used to be at USC) demonstrated a complementary reaction to Gassman’s reaction above; when Et2AlCl is used as the Lewis acid instead of ZnI2nitriles are obtained instead. The mechanism invoked by Prof. Weber involves a little more hand-waving, however:

cyanide2

The first step involves the interconversion of TMSCN with its isocyano isomer. It’s not far-fetched on paper, and you can certainly defend this using the Curtin-Hammett principle. However, the literature support for this is rather weak; detailed spectroscopic studies of triorganosilyl cyanides gave no evidence for the presence of the isocyano form. However, another Japanese group studied this set of reactions with more Lewis Acids, and what seems apparent to me is that soft Lewis acids seem to promote formation of the isocyanide, whereas hard Lewis acids promote formation of the cyanide. Thus, two different mechanisms are at play depending on the Lewis acid involved. With reactions involving cyanide, the nitrogen preferentially attacks hard electrophiles (i.e. carbocations, giving the Ritter reactions, as well as other electron-deficient species). My proposal is that the first step would be a nitrilium ion formed from TMSCN attacking the aluminum atom; this species would be the active cyanating agent. If anyone is up for it, it may be possible to characterize this species; Melanie Sanford recently wrote about rapid-injection (RI)-NMR being used to characterize transient Cu(III) intermediates, and the same technique could possibly be used here. In contrast, other Lewis acids (ZnI2, Pd salts, etc.) activate the epoxide for nucleophilic ring-opening by attack of the nitrogen in TMSCN, which is drawn to the nascent carbocation by Coulombic forces.

do have some ideas for new synthetic reactions based on this chemistry, but that will have to be explored once I can get back in a lab.

References:

  1. Gassman, P. G.; Guggenheim, T. L. J. Am. Chem. Soc. 1982104, 5849 
  2. Spessard, G. O.; Ritter, A. R.; Johnson, D. M.; Montgomery, A. M. Tetrahedron Lett. 198324, 655 (This paper was published independently and at the same time as Gassman’s paper above, and describes the same results)
  3. Gassman, P. G.; Talley, J. J. Org. Synth. 198160, 14  (Gassman’s 1981 prep for converting aldehydes/ketones to cyanohydrins with TMSCN)
  4. Okada, I.; Kitano, Y. Synthesis 201124, 3997 (Refs. 3-9 cover converting various functional groups to isocyanides)
  5. Kitano, Y; Chiba, K.; Tada, M. Tetrahedron Lett. 199839, 1911
  6. Kitano, Y.; Chiba, K. Tada, M. Synthesis20013, 437
  7. Kitano, Y.; Chiba, K. Tada, M. Synlett19993, 288
  8. Kitano, Y.; Manoda, T.; Miura, T.; Chiba, K.; Tada, M. Synthesis 20063, 405
  9. Pronin, S. V.; Reiher, C. A.; Shenvi, R. A. Nature 2013501, 195
  10. Mullis, J. C.; Weber, W. P. J. Org. Chem. 1982 47, 2873 (Weber’s conditions for the ring-opening of epoxides and oxetanes with TMSCN + Et2AlCl)
  11. Seckar, J. A.; Thayer, J. S. Inorg. Chem. 1976 15, 501 (Detailed spectroscopic study on the interconversion of the iso- and normal forms of triorganosilyl cyanides)
  12. Hickman, A. J.; Sanford, M. S. Nature2012484, 177 (Review in which various methods for characterizing transient high-valent metal intermediates are discussed, including RI-NMR)

This is by no means an exhaustive list; I have many more papers with me on this topic. If you want them, let me know.

Finally, I have to include this link to Prof. Andrei Yudin’s blog, which got this whole discussion started in my mind.

Blog at WordPress.com.