feat: netmix2 by tristan-f-r · Pull Request #225 · Reed-CompBio/spras

tristan-f-r · 2025-05-23T18:07:55Z

Closes #157. We need a SPRAS gurobi server license. Once this is integrated, this lays the ground for #105.

ntalluri · 2025-05-27T16:28:25Z

+        output_vertices.rename(output_file)
+
+    @staticmethod
+    def parse_output(raw_pathway_file, standardized_pathway_file):


Netmix2 will not work with the ML post processing and possibly some of the other post processing. I'm not sure how to fix this at the moment, but this is something that will need to be considered.

Is there something bad about getting the induced subgraph that makes netmix2/RWR not work well with ML post processing?

a. a 1 month mistake

ntalluri · 2026-04-27T15:54:07Z

This PR doesn't have the pydantic style. Reminder to update this PR.

ntalluri · 2026-05-11T17:59:28Z

      run1:
        n: 1

+  - name: "netmix2"


can we add some defaults here?

ntalluri · 2026-05-11T18:03:12Z

getting this error when I try to run this locally:

ImportError in file "/Users/nehatalluri/Desktop/research/spras/Snakefile", line 2:
cannot import name 'algorithms' from partially initialized module 'spras.runner' (most likely due to a circular import) (/Users/nehatalluri/Desktop/research/spras/spras/runner.py)
  File "/Users/nehatalluri/Desktop/research/spras/Snakefile", line 2, in <module>
  File "/Users/nehatalluri/Desktop/research/spras/spras/runner.py", line 11, in <module>
  File "/Users/nehatalluri/Desktop/research/spras/spras/netmix2.py", line 13, in <module>
  File "/Users/nehatalluri/Desktop/research/spras/spras/secrets.py", line 9, in <module>
  File "/Users/nehatalluri/Desktop/research/spras/spras/config/config.py", line 27, in <module>
  File "/Users/nehatalluri/Desktop/research/spras/spras/config/schema.py", line 17, in <module>
  File "/Users/nehatalluri/Desktop/research/spras/spras/config/algorithms.py", line 20, in <module>

akniyetnurbol

Comments describe the changes made to get NetMix2 running locally through SPRAS and the tests passing. Also, includes a note on how the Gurobi license is configured in config.yaml and accessed via secrets.py.

akniyetnurbol · 2026-06-30T18:45:31Z

+            node_df = data.get_node_columns(['prize'])
+        else:
+            raise ValueError("Node prizes are required for NetMix2.")
+        node_df.to_csv(filename_map['scores'], index=False, columns=['prize', 'NODEID'], header=False, sep='\t')


The order of the column should be changed, so it's node then prize, but also raw SPRAS prizes are random weights and can't be used as p-values, so np.exp(-prize) is needed to convert them to valid numbers from 0 to 1 as required by locfdr

import numpy as np
node_df['pval'] = np.exp(-node_df['prize'])
node_df.to_csv(filename_map['scores'], index=False, columns=['NODEID', 'pval'], header=False, sep='\t')

I think we might want to add np.exp(-prize) as its own function in the same place we have the -log(p-val) function - that would keep these consistent since they transform scores into p-values and vice versa.

Also, this doesn't exactly transform scores into p-values because there's no statistical test here - it's just converting them to numbers between 0 and 1 where lower is better.

Does Netmix require the values to be between 0 - 1 or is this a statistical test that the Netmix2 paper used to convert their data?

By my understanding, it's less of a statistical choice and more of a hard requirement in the code based on how NetMix2 processes inputs. It converts scores to z-scores using norm.ppf, which only works for values in (0, 1], so the input just needs to be a number in that range where lower is better.

akniyetnurbol · 2026-06-30T18:55:08Z

+        bind_path, scores_file = prepare_volume(inputs["scores"], work_dir, container_settings)
+        volumes.append(bind_path)
+
+        bind_path, license_file = prepare_volume(inputs["gurobi_path"], work_dir, container_settings)


the gurobi path is not in the inputs dict, so it causes an error when run() is used. Could use the gurobi_path variable from line 52

bind_path, scores_file = prepare_volume(gurobi_path, work_dir, container_settings)

akniyetnurbol · 2026-06-30T20:01:59Z

+        edges_df.to_csv(filename_map['network'], index=False, sep='\t', columns=['Interactor1', 'Interactor2'], header=False)
+
+    @staticmethod
+    def run(inputs, output_file, args, container_settings=None):


Check for no input cases in this function

akniyetnurbol · 2026-06-30T20:14:59Z

+
+import pytest
+
+import spras.config as config


import spras.config.config as config since init_from_file lives in spras/config/config.py

import spras.config.config as config

akniyetnurbol · 2026-06-30T20:22:26Z

+class TestNetMix2:
+    def test_nm2_required(self):
+        OUT_FILE.unlink(missing_ok=True)
+        NetMix2.run(


Inputs must be a dict, not keyword args (consistent with other SPRAS wrappers)
Also, if the container is set to None, then prepare_volume raises an error, so container_settings should be passed explicitly

NetMix2.run(
{"network": TEST_DIR / 'input' / 'network-basic.txt',
"scores": TEST_DIR / 'input' / 'scores-basic.txt'},
output_file=OUT_FILE,
args=NetMix2Params(num_edges=1),
container_settings=config.config.container_settings
)

akniyetnurbol · 2026-06-30T20:54:44Z

+pandas==1.1.5
+statsmodels==0.12.2
+gurobipy==9.5.2
+networkx==2.5.1


Should add packaging for patsy with a consistent version

packaging==24.0

akniyetnurbol · 2026-06-30T21:03:19Z

+
+    def test_nm2_missing(self):
+        with pytest.raises(ValueError):
+            NetMix2.run(


TypeError because args was missing

NetMix2.run(
{"network": TEST_DIR / 'input' / 'network-basic.txt'},
output_file=OUT_FILE,
args=NetMix2Params(),
container_settings=config.config.container_settings #config.container_settings
)

akniyetnurbol · 2026-06-30T21:56:17Z

+
+    @staticmethod
+    def run(inputs, output_file, args, container_settings=None):
+        gurobi_path = gurobi()


gurobi() function in spras/secrets.py is used to get the Gurobi license path, which is read from secrets.gurobi in config.yaml

When the user downloads the gurobi.lic file, they need to update the directory in config.yaml under "secrets:"
OR
Place it at the location already indicated

annaritz · 2026-07-01T21:05:39Z

Proposed edits look good, we'll check with @tristan-f-r to see whether you should do a PR to tristan-f-r-:netmix2 to get the changes in.

ntalluri · 2026-07-10T17:15:04Z

@akniyetnurbol I saw your email about the pval - prize prize - pval conversion. Would you be able to explain why in SPRAS we would need this conversion for Netmix2? In my opinion pval - prize conversions would happen before putting data into SPRAS.

akniyetnurbol · 2026-07-10T18:23:35Z

@akniyetnurbol I saw your email about the pval - prize prize - pval conversion. Would you be able to explain why in SPRAS we would need this conversion for Netmix2? In my opinion pval - prize conversions would happen before putting data into SPRAS.

@ntalluri Yeah, when I was running snakemake there was an error because NetMix2 explicitly expects p-values as input (per its readme and paper), but spras stores node data as prizes using -log10(p-value). So by the time data enters spras it's already been converted from p-values to prizes. The generate_inputs wrapper converts back using 10^(-prize), which recovers the original p-value. I put it in the wrapper so the user doesn't have to handle this themselves just to keep things consistent with other algorithms in spras.

NetMix2 changes

annaritz · 2026-07-28T23:02:11Z

I think we are nearly ready to review this PR, after @akniyetnurbol's changes. I added the gurobi license to this repo as a secret.

There seems to be two unresolved conflicts in spras/runner.py related to how algorithms are loaded - @akniyetnurbol or @tristan-f-r can you resolve those? The remaining tests will trigger once the conflict is resolved.

annaritz · 2026-07-29T18:13:46Z

In talking with @akniyetnurbol, it seems like we might need to set a GitHub Environment for the gurobi license to work. Will wait to see if the tests fail before adding that.

# Conflicts: # spras/runner.py

Netmix2 conflict resolved

tristan-f-r added 2 commits May 23, 2025 18:07

feat: begin netmix2

267dcb8

feat: properly parse input

101c35f

tristan-f-r commented May 23, 2025

View reviewed changes

Comment thread config/config.yaml Outdated

tristan-f-r added 5 commits May 23, 2025 20:09

get vertex list

604d743

feat: options, begin scaffolding test

20dfa35

test: init

3c830f2

fix: typos, finish docker container, begin testing

832bec3

fix: flags, finally a license error!

71a84ab

tristan-f-r changed the title ~~feat: begin netmix2~~ feat: netmix2 May 23, 2025

tristan-f-r mentioned this pull request May 23, 2025

NetMix2 #157

Open

ntalluri reviewed May 27, 2025

View reviewed changes

tristan-f-r added 2 commits May 27, 2025 18:01

feat: begin secrets

653bc7c

chore: todo cmt on env

a7ac3ae

tristan-f-r added the algorithm Pathway reconstruction algorithm to add to SPRAS label May 29, 2025

tristan-f-r mentioned this pull request Jun 18, 2025

feat: allow multiple env variables #273

Merged

tristan-f-r added the blocked-by-other-pr label Jun 26, 2025

tristan-f-r removed the blocked-by-other-pr label Jul 7, 2025

Merge branch 'umain' into netmix2

0903200

tristan-f-r mentioned this pull request Jul 7, 2025

Generic spras limit option on containers #316

Open

tristan-f-r added 7 commits July 7, 2025 22:13

chore: begin addressing cmts

c236e2f

feat: proper .env file parsing

df094c7

fix: better cmt handling

725474f

fix: add input network

291299e

fix: don't parse as env

9b5cfb6

a. a 1 month mistake

test: on nm2

e9b5d2b

test: netmix2 parse outputs

99d299c

tristan-f-r marked this pull request as ready for review July 7, 2025 23:12

tristan-f-r mentioned this pull request Jul 11, 2025

fix: let raw_pathway_df handle present headers but no data #236

Merged

github-actions Bot added the merge-conflict This PR has merge conflicts. label Oct 3, 2025

Merge branch 'main' into netmix2

179b8f2

tristan-f-r added the awaiting-author Author of the PR needs to fix something from a review / etc. label Jan 23, 2026

Merge branch 'umain' into netmix2

cfd7a7f

github-actions Bot added merge-conflict This PR has merge conflicts. and removed merge-conflict This PR has merge conflicts. labels Mar 16, 2026

Merge remote-tracking branch 'upstream/main' into netmix2

8733a63

github-actions Bot removed the merge-conflict This PR has merge conflicts. label Apr 28, 2026

refactor: netmix2

a9c32eb

ntalluri reviewed May 11, 2026

View reviewed changes

Comment thread config/config.yaml

run1:

n: 1

- name: "netmix2"

ntalluri May 11, 2026

Copy link
Copy Markdown

Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can we add some defaults here?

github-actions Bot added the merge-conflict This PR has merge conflicts. label Jun 6, 2026

akniyetnurbol suggested changes Jun 30, 2026

View reviewed changes

akniyetnurbol added 3 commits July 9, 2026 06:04

NetMix2 changes

b9e3172

Fix yaml indentation in config

d47448d

Fix singularity test to use dict input format

5304c3d

akniyetnurbol and others added 7 commits July 22, 2026 10:57

docker cython

4c4c726

quick lic check

4546362

license workflow

a9c6781

trigger CI on netmix2 branch

222a6a5

running tests with docker

b1c14b1

pre commit check errors and config file

4861470

Merge pull request #3 from akniyetnurbol/akniyet-netmix2-fixes

5c3d829

NetMix2 changes

akniyetnurbol and others added 2 commits July 29, 2026 15:46

Merge remote-tracking branch 'upstream/main' into tristan-netmix2

1419844

# Conflicts: # spras/runner.py

Merge pull request #4 from akniyetnurbol/netmix2-conflict-resolved

285ac84

Netmix2 conflict resolved

github-actions Bot removed the merge-conflict This PR has merge conflicts. label Jul 30, 2026

Uh oh!

Conversation

tristan-f-r commented May 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ntalluri commented Apr 27, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ntalluri commented May 11, 2026

Uh oh!

akniyetnurbol left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

annaritz commented Jul 1, 2026

Uh oh!

ntalluri commented Jul 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

akniyetnurbol commented Jul 10, 2026

Uh oh!

annaritz commented Jul 28, 2026

Uh oh!

annaritz commented Jul 29, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

tristan-f-r commented May 23, 2025 •

edited

Loading

ntalluri commented Jul 10, 2026 •

edited

Loading