Who works on what in QCB

USC Quantitative and Computational Biology faculty, placed by what their papers are about. Made by students in QBA, not by USC, from OpenAlex papers as of fall 2026.

View
Arrange faculty

Nearby means similar research, though distances between far-apart groups don't mean much. Bigger marks are bigger labs.

How this was made

Most of this comes from what faculty have published. The technique view also reads labs' own websites and adds what students who worked there have told us.

Papers

Each person on the QCB faculty page was matched by hand to their records in OpenAlex, a free catalog of scholarly papers. Peer reviews, corrections and supplementary files were dropped, along with papers by other people who share their name. That left papers.

Research similarity

A language model trained on scientific papers (SPECTER2) turned each paper's title and abstract into a list of 768 numbers, placing similar papers close together. A person's position is the average of their papers. Recent papers count more, as do papers where they're first or last author.

The map

t-SNE and UMAP squeeze those 768 dimensions into two. Both keep each person near their closest research neighbors, at the cost of distances between distant groups, which don't mean much. The two views are different summaries of the same data, so trust what they agree on.

3D

The 3D view runs the same method on the same numbers with a third dimension, so less has to be squeezed. On every map it scores a little better on trustworthiness, a check that people who look close really are: 0.96 against 0.94 for t-SNE on the QCB map. It's still a summary. Marks in front are bigger and stand out more, and the spin itself means nothing.

Constellation draws the same 3D map as a night sky. Faculty are the stars, at the same positions, with size still showing the lab and color the cluster, though shape no longer shows someone's role. The lines are still shared papers, curved over the sphere. The comets on them and the faint stars behind them mean nothing.

Clusters

Average-linkage hierarchical clustering on the same numbers; the Tree view shows the full hierarchy. The cluster names were written by hand.

By technique

By field arranges people by what they study. By technique arranges them by how they work, using a hand-written list of quantitative and computational methods, matched against each lab's paper titles, abstracts and keywords, and against its own website and faculty profile where those could be read ( of labs). The technique positions and clusters use the same t-SNE, UMAP, MDS and clustering steps as the field view, on those method profiles. Each method appears at one level for a lab:

Demonstrated: the lab does this work, by at least 3 papers making up at least 4% of its weighted papers, or two separate statements on its website. Firsthand: a student who worked there told us, which papers can't show. Potential: the lab generates data this method could use, or mentions the method once. A microscopy lab has computer vision potential even if it has never published on computer vision.

This is word matching, so labels can be wrong or missing, and labs that don't name their methods in abstracts look less quantitative than they are. Open a label to see the paper or source behind it, so you can check it yourself. In a spot check, about 9 in 10 labels that came from websites alone were right.

Related fields

Expanding adds faculty from fields QBIO students also do research in: Biostatistics, Biomedical Engineering, Molecular and Cellular Biosciences, Marine and Environmental Biology, Neurobiology and the Brain and Creativity Institute, plus a few labs students have worked in. That makes papers in all. Everyone's position is recalculated, so QCB faculty move too, and MDS replaces UMAP, which keeps distances between far-apart groups more honest. By field opens on MDS, with t-SNE one click away. of the added labs have sizes counted from their websites. The other have no count, so their marks are dashed and drawn at a typical size of .

Lines

Two people are joined if they share a paper. Each shared paper adds 1 ÷ (authors − 1) to the pair's strength, so a two-author paper counts 1 and a 100-author consortium paper about 0.01.

Finding your match

The model above is too big to run in your browser, so it has already read about 66,000 research words and phrases, from these papers and from OpenAlex's list of keywords. Each one comes with a score for every faculty member, from how close its meaning is to their papers and how much of their work uses it. Your text is split into the longest terms it can find, so “protein engineering” stays one idea instead of “protein” plus “engineering”. If a phrase isn't on the list, the panel says its words were matched separately. The panel shows the topics it read and the words it didn't use, like ones too general to tell people apart. Leaving a topic out lowers each person by how far above average their papers use it, so people known for it move down the list. When no one stands out by more than about 1.5 standard deviations, or half of what you typed isn't in these papers, the panel says the match is weak. Nothing you type leaves your browser. Given only the title of a paper it hadn't seen, this finds the real author in the top 5 about 93% of the time (88% with related fields added, where there are more people to choose from), but your own words will be vaguer than a title. Your dot sits at the weighted average of your closest matches, which is an estimate: t-SNE and UMAP have no exact way to add a new point. A thrown dot works in reverse, blending the people nearest where it lands.

Lab sizes

A mark's area shows the size of the lab. For labs that's everyone the lab's own website lists, including the PI and any undergraduates. Labs differ in who they list and how often they update, so compare these loosely. The other are estimates: USC co-authors on the person's senior-author papers since 2023, plus anyone QCB lists in the lab, plus the PI. Checked against the labs we could count, these are about right on average, but about a quarter are off by more than a factor of two. faculty have no count and too few recent USC papers to estimate, so their marks are dashed and drawn at a typical size of .

Stats

The h-index, citations and papers per year come from the same OpenAlex papers as the map. Early work in another field gets filtered out along with other people's papers, so these numbers can run lower than OpenAlex's own profile pages, and usually lower than Google Scholar's.

Grants

Grants come from NIH RePORTER and NSF's award search, checked on . Both are searched by name, so each person's NIH and NSF records were matched by hand, leaving out other people with the same name. A grant is one NIH project, with its yearly renewals and any projects inside a larger center grant counted once, or one NSF award. It counts as active until its end date. When a grant was awarded to another institution, such as one from before someone came to USC or one led from another university, it says which. Foundation, Department of Defense and subaward funding aren't in these sources, so a lab can have more funding than is listed. Dollar amounts are left out, because a multi-PI or center grant's total says little about one lab. Grants are listed for the QCB faculty only, not for related fields.

PhD students

The QCB website lists PhD students in labs. more haven't joined a lab yet, and students from other programs aren't listed, so a lab can have more students than you see here. Common names make papers hard to match, so a student's OpenAlex profile only counts if it appears on a paper by their advisor or has been checked by hand, and only papers listing USC or a QCB faculty co-author are shown. Some students will have papers that aren't found.

Saved faculty

Your saved list is kept in this browser on this device, and nothing about it is sent anywhere. Clearing your browser's site data removes it, so copy the list to keep it somewhere else. Emails come from department websites, and some are missing.

What it can't tell you

It only sees published work and what labs post online, so new collaborations and unpublished projects won't appear, and some papers will be missing or misattributed. It says nothing about who is taking students. Check with faculty directly.

Visit counts

Visits to this page are counted with GoatCounter, which uses no cookies and collects no personal data. What you type never leaves your browser.