It's hard for me to type, because I'm traveling with no laptop, so I decided to leave a video review of the in progress manuscript @zietzm printed for me. It's split into three videos available on the following album https://photos.app.goo.gl/VYuWFFMMVCXBU47c8

As always, feel free to disregard any advice that you disagree with. I think the manuscript has improved a lot, and with a bit more focus on nailing home the main points, this will be a strong study on a challenging topic!

Second, we assumed that the XSwap process is stationary. This assumption also does not actually hold, but it was made because it significantly simplifies the problem.

@zietzm Why do we say the xswap process isn't stationary? This is not obvious to me.

Good news in email from 2023-11-01:

GIGA-D-23-00002R1
The probability of edge existence due to node degree: a baseline for network-based predictions
Michael Zietz; Daniel S. Himmelstein; Kyle Kloster; Christopher Williams; Michael W. Nagle; Casey S. Greene
GigaScience

Dear Prof. Greene,

Appreciate your patience.

Peer review for your manuscript has been completed and your paper is provisionally accepted pending any final minor changes, please work with the curators to prepare a GigaDB DOI for your datasets, as this will need to be cited in the paper before we begin the production process.

Our curators have been informed and they will contact you within a week.

Please note we cannot formally accept your manuscript (and send it to production) until the GigaDB DOI citation is finalized and added to the manuscript.

Warm regards,
Hongfang

I'll note some things here that we'll want to do before the final DOCX submission to GigaScience based on what we learned from the connectivity-search-manuscript proofing process.

We've submitted this manuscript to GigaScience (GIGA-D-23-00002). Part of the submission process is an editorial check to ensure data availability. Super excited to see a journal taking such proactive measures.

The check identified some data that we need to better document, export, and upload to GigaDB. Quoting from an email from Yannan Fan on 2023-01-11:

Dear Casey S. Greene,

At GigaScience Press we have strict transparency and reproducibility policies; to ensure this is maintained it is imperative that all data and tools required to reproduce your manuscript are available to the reviewers and ultimately the readers. To facilitate this we have developed GigaDB, which is the repository used to host data and tools associated with articles in GigaScience and GigaByte. You can see a complete workflow of the combined submission process here. This email represents the step labeled as "provide authors with private FTP login".

I have had a quick read through your manuscript. To enable full reproducibility and transparency please follow these instructions on how to make your data available:

• Please read the guidelines for submission of software data: http://gigadb.org/site/guidesoftware

• You should include a list of md5sum values for all files uploaded so that we can confirm file integrity after transfer.

• Please also include a readme.txt file listing all files with a 1 line description of each.

  • Please provide the raw data underlying Figure 4A-B, Figure 5, Figure 6A-C, Figure 7, and Figure 9.

I've opened this issue so we can coordinate and track progress fulfilling this data request.

1765_001.pdf

Old-school markup attached, that should probably be broken into multiple issues. Mixture of fine-grained (e.g. typos and concerns about figure legends/clarity) and broader-scale concerns (e.g. lack of definitions of technical terms, organization, and seemingly overly-explanatory language "justifying" the approach/results).

Overall, I have some qualms about the tone of the paper (e.g. what is being explained, where, and at what level of detail), and would be happy to have a call to discuss this more broadly. I also feel like the paper suffers from a lack of technical specificity in much of the narrative. I'm still not 100% sure I know what you algorithm you're using when you say "edge reconstruction", which worries me.

Let me know how you'd like to approach things from here. Thanks!

This issue is for commenting with relevant literature or prior work. For substantive discussion about a prior work, consider opening a new issue.

Since the edge prior is based only on degree, it is unsurprising that it exhibits weak performance in predicting a network with a different degree distribution.
We have considered the edge prior as a baseline edge predictor, whose performance indicates the utility of degree for a specific prediction task.
The edge prior's low performance in the third task indicates that degree is less helpful for edge prediction tasks in which training and testing networks do not share their degree distributions.
Moreover, we believe such between-distribution prediction may be a relatively common task, with examples given by the networks in Figure {@fig:degree-bias}.

I feel like this paragraph could use some tuning. It would help to give the take-home message more clearly to aid readers. I think that it's there, but it's hidden a bit between the lines.

Manubot is out of date.

Also seems that we're building on both:

• https://travis-ci.org/greenelab/xswap-manuscript
• https://travis-ci.com/greenelab/xswap-manuscript

Ideally we should migrate to travis-ci.com and also consider switching the the encrypted variable SSH key method.

@zietzm is currently working in the Greene Lab for the summer before he departs to start his PhD at Columbia University in the fall. @zietzm's work from his time in the lab will likely be divided into two manuscripts:

1. This manuscript in greenelab/xswap-manuscript, viewable here, which covers creating randomized networks via the XSwap permutation algorithm to calculate a prior probability of edge existence based on node degree (termed the "edge prior"). Repositories related to this work include https://github.com/greenelab/xswap-analysis, https://github.com/hetio/xswap, and discussion in greenelab/connectivity-search-analyses#134.

2. The hetnet connectivity search manuscript regarding the webapp at https://het.io/search/. This is a separate manuscript which we hope to begin writing soon. This will cover work like matrix DWPC computation, DWPC null distributions, etcetera. We will address authorship for this work separately.

@cgreene, @zietzm, and I discussed contributors to this manuscript and have identified the involvement of the following individuals (in provisional authorship order):

1. @zietzm
2. @dhimmel
3. @kkloste -- deriving an analytic method for the edge prior
4. @chrsunwil -- assisting with a C implementation of xswap
5. @naglem -- discussions of relevance to applications
6. @bdsullivan -- participation in site visit and phone discussion
7. @cgreene

If you are listed above and would like to be a coauthor, please open a pull request to add your information by going here. Adding yourself as an author indicates that you agree to ICMJE author guidelines and will stay involved in reviewing and contributing to the manuscript as needed. Finally, author order and membership is provisional and may be reevaluated at any time to account for the latest contributions.

I have noticed as I'm reading that we use "picked up on" a fair bit. I don't think that is going to be formal enough for academic writing. We should look for instances of this after we merge things and replace them.

Hello Manubot user.

If you were not already aware, your paper is being showcased in the Manubot Catalog.

Previously, we were making our own thumbnails for each paper, and hosting them in the same repository as manubot.org. But now we are moving to a system of having each paper store its thumbnail in its own repository, and linking to it from there.

Here is our suggested thumbnail for your paper that follows the new thumbnail guidelines:

Please place this image in the root of your repository with the name thumbnail.png. Once you have, let us know in this issue and we'll do the rest to update the catalog.