Details

    • Type: Task
    • Status: Closed (View Workflow)
    • Priority: Major
    • Resolution: Done
    • Affects Version/s: None
    • Fix Version/s: None
    • Labels:
      None
    • Story Points:
      0.5
    • Sprint:
      Spring 2 2022 Jan 18 - Jan 28, Spring 3 2022 Jan 31 - Feb 11, Spring 4 2022 Feb 14 - Feb 25

      Description

      During a live demo, one of the attendees noticed a problem with a graph operation.

      The sum of two graphs did not appear to be correct.

      To repeat / investigate:

      • Add data source from IGBF-3064
      • Load six scaled coverage graphs from the graphs folder
      • Select 3 of the graphs by shift-clicking the left side graph track labels (red outline appears around the label when the graph track is selected)
      • Select use the Graph Operations multi-graph menu "Mean" to create a new graph, which is supposed to be the average of the selected graphs
      • Repeat the above with another 3 graphs
      • Now, subtract the mean graphs from each other
      • Check the y-axis values carefully to ensure that the subtraction was correctly done.

      Report what you see, along with images showing incorrect or correct behavior.

        Attachments

        1. 48738118.png
          48738118.png
          265 kB
        2. Difference.png
          Difference.png
          101 kB
        3. img1.png
          img1.png
          289 kB
        4. img2.png
          img2.png
          286 kB
        5. img3.png
          img3.png
          277 kB
        6. img4.png
          img4.png
          257 kB
        7. Mean  1.png
          Mean 1.png
          93 kB
        8. Mean 2.png
          Mean 2.png
          100 kB

          Issue Links

            Activity

            Hide
            nfreese Nowlan Freese added a comment - - edited

            Additional testing:

            Add http://lorainelab-quickload.scidas.org/hotpollen/ as a new data source.
            Open the S_lycopersicum_Sep_2019 genome
            Navigate to SL4.0ch02:48,738,056-48,738,176

            Add the following scaled coverage graphs:
            are 28 only, rep 1, scaled coverage
            are 28 only, rep 2, scaled coverage
            are 28 only, rep 3, scaled coverage
            are 28 to 34, rep 1, scaled coverage
            are 28 to 34, rep 2, scaled coverage
            are 28 to 34, rep 3, scaled coverage

            Click Load Data
            Create Mean graphs for the two groups of 3
            Hide the original 6 scaled coverage graphs
            Create Diff graph from the means
            Navigate to SL4.0ch02:48,737,813-48,738,983
            Click Load Data 3 times

            This appears to break the Diff graph as it should now show incorrect data.

            If all data including mean and diff graphs are loaded at SL4.0ch02:48,737,813-48,738,983 they appear correctly (see img1.png). It is only when the view is zoomed in initially, the data loaded, and then zoomed out that the diff graph appears to become corrupted (see img2.png). From examining the y-coord in the corrupted diff track, it appears that the data are a copy of the bronze colored mean track (see img3.png). IGB also appears to have trouble rendering the track, which is why it appears on both the positive and negative axis in img2.png.

            From this initial testing I can make a few guesses as to what might be happening. When IGB creates tracks based on multi-graph operators it first needs to load the data before calculating the multi-graph operator track. From the user's perspective they may need to click the Load Data button multiple times. In this example, there are two multi-graph operator tracks, so IGB needs to follow the correct order of operations to successfully create the diff of the means track. When I clicked load data it appeared that only one of the mean tracks was created successfully while the other was in limbo (see img4.png). Clicking load data again caused the corrupt diff track to appear.

            In addition, once the corrupt diff track has appeared the track is broken from that point forward. Removing and reloading the track does not appear to fix the issue, only restarting IGB can correct the track. My guess would be that something is incorrectly being included in a data model for that track in IGB and is causing any attempt to further load data in the diff track to fail. However, nothing is appearing in the IGB logs.

            While this is somewhat of an edge case I think it is worth further investigation as it could be very confusing for a user.

            Tested on Mac 11.6.3 on IGB 9.1.8 release.

            Show
            nfreese Nowlan Freese added a comment - - edited Additional testing: Add http://lorainelab-quickload.scidas.org/hotpollen/ as a new data source. Open the S_lycopersicum_Sep_2019 genome Navigate to SL4.0ch02:48,738,056-48,738,176 Add the following scaled coverage graphs: are 28 only, rep 1, scaled coverage are 28 only, rep 2, scaled coverage are 28 only, rep 3, scaled coverage are 28 to 34, rep 1, scaled coverage are 28 to 34, rep 2, scaled coverage are 28 to 34, rep 3, scaled coverage Click Load Data Create Mean graphs for the two groups of 3 Hide the original 6 scaled coverage graphs Create Diff graph from the means Navigate to SL4.0ch02:48,737,813-48,738,983 Click Load Data 3 times This appears to break the Diff graph as it should now show incorrect data. If all data including mean and diff graphs are loaded at SL4.0ch02:48,737,813-48,738,983 they appear correctly (see img1.png). It is only when the view is zoomed in initially, the data loaded, and then zoomed out that the diff graph appears to become corrupted (see img2.png). From examining the y-coord in the corrupted diff track, it appears that the data are a copy of the bronze colored mean track (see img3.png). IGB also appears to have trouble rendering the track, which is why it appears on both the positive and negative axis in img2.png. From this initial testing I can make a few guesses as to what might be happening. When IGB creates tracks based on multi-graph operators it first needs to load the data before calculating the multi-graph operator track. From the user's perspective they may need to click the Load Data button multiple times. In this example, there are two multi-graph operator tracks, so IGB needs to follow the correct order of operations to successfully create the diff of the means track. When I clicked load data it appeared that only one of the mean tracks was created successfully while the other was in limbo (see img4.png). Clicking load data again caused the corrupt diff track to appear. In addition, once the corrupt diff track has appeared the track is broken from that point forward. Removing and reloading the track does not appear to fix the issue, only restarting IGB can correct the track. My guess would be that something is incorrectly being included in a data model for that track in IGB and is causing any attempt to further load data in the diff track to fail. However, nothing is appearing in the IGB logs. While this is somewhat of an edge case I think it is worth further investigation as it could be very confusing for a user. Tested on Mac 11.6.3 on IGB 9.1.8 release.
            Hide
            nfreese Nowlan Freese added a comment -

            I think the next steps for this issue would be for someone to follow the steps outlined above to replicate the issue.

            If the issue does appear to be repeatable, then I would suggest we create a new ticket to attempt to resolve the issue.

            Show
            nfreese Nowlan Freese added a comment - I think the next steps for this issue would be for someone to follow the steps outlined above to replicate the issue. If the issue does appear to be repeatable, then I would suggest we create a new ticket to attempt to resolve the issue.
            Hide
            omarne Omkar Marne (Inactive) added a comment -

            Followed above testing instructions on IGB 9.1.10 with the build date/time : 01-24-2022 15:26:44 and the latest commit id : 6015d4c2488ec0abab95bd36b2ff1d94b3c3f9be and Operating System - Windows 11.

            The issue does appear repeatable.

            Dr. Freese is creating a new ticket to resolve the issue.

            Show
            omarne Omkar Marne (Inactive) added a comment - Followed above testing instructions on IGB 9.1.10 with the build date/time : 01-24-2022 15:26:44 and the latest commit id : 6015d4c2488ec0abab95bd36b2ff1d94b3c3f9be and Operating System - Windows 11. The issue does appear repeatable. Dr. Freese is creating a new ticket to resolve the issue.
            Hide
            ann.loraine Ann Loraine added a comment -

            Thank you Omkar Marne and Nowlan Freese. Moving to Done, with the expectation that Nowlan Freese will create a new issue to address this.

            I think it is indeed an important issue to address because there is now way for a user to understand the sequence of events that need to be performed in order to achieve the correct result. Also, this issue may have been noticed prior to this. It would be worthwhile to search Jira for prior observations of this same problem.

            Show
            ann.loraine Ann Loraine added a comment - Thank you Omkar Marne and Nowlan Freese . Moving to Done, with the expectation that Nowlan Freese will create a new issue to address this. I think it is indeed an important issue to address because there is now way for a user to understand the sequence of events that need to be performed in order to achieve the correct result. Also, this issue may have been noticed prior to this. It would be worthwhile to search Jira for prior observations of this same problem.
            Hide
            nfreese Nowlan Freese added a comment -

            I have created IGBF-3084 to further investigate this issue.

            Show
            nfreese Nowlan Freese added a comment - I have created IGBF-3084 to further investigate this issue.

              People

              • Assignee:
                omarne Omkar Marne (Inactive)
                Reporter:
                ann.loraine Ann Loraine
              • Votes:
                0 Vote for this issue
                Watchers:
                3 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: